Re: [sage-devel] Learning from past experience
William Stein
<bjarke...@gmail.com> wrote:
>
> People have been writing mathematics software for a long time and so
> people have been accumulating experiences about good ways and bad ways
> to do things. I'm wondering how much Sage development is informed by
> that experience and however much it is, I think it is worthwhile to
> think about how to increase that. Is this something that has been
> discussed here?
>
> Given some feature, how to find out what experiences other people have
> had in implementing it before? One way is to have the people with
> those experiences directly involved in developing Sage or at least
> have them on sage-devel. A second is to identify these people and ask
> them what they think. A third is to examine other systems and see what
> kinds of advantages and disadvantages their designs entail (open
> source systems allows to read the code, too). A fourth is to search
> the literature for anything relevant. An with all of these except that
> first is that they delay development by adding steps beyond coding the
> first seemingly reasonable thing that comes to mind. That is both good
> and bad.
>
> I thought to open this topic here because of a recent discussion on
> sage-flame, and I'm hesitant to do so since it has the potential to
> generate lots of talk to no benefit. On the other hand, it seems to me
> that if a good way can be found to benefit even a little more from
> past experience, that could have a significant impact on the quality
> of Sage. My impression is that asking the people with the relevant
> experience rarely happens, unless those people are already on sage-
> devel and choose to speak up, and doing more of that seems a low-
> effort way to benefit a little more from past experience.
I suspect that this discussion should be on sage-flame, but we'll see.
Your disguising a dubious claim in the form of a question: Since Sage
developers just do the first thing that pops into their head, how can
we find a way to fix this problem? However, implicit in the question
is the assertion that people who implement code for Sage often do the
"first seemingly reasonable thing that comes to mind" instead of
"asking the people with the relevant experience". This strikes me as
a naive and generalization about a subtle and complicated development
process.
I've cc'd this to sage-flame. I recommend that all responses be there only.
William
Tom Boothby
<bjarke...@gmail.com> wrote:
>
> People have been writing mathematics software for a long time and so
> people have been accumulating experiences about good ways and bad ways
> to do things. I'm wondering how much Sage development is informed by
> that experience and however much it is, I think it is worthwhile to
> think about how to increase that. Is this something that has been
> discussed here?
It's been discussed to death. Predominantly by one Richard Fateman
who will remain unnamed.
> Given some feature, how to find out what experiences other people have
> had in implementing it before? One way is to have the people with
> those experiences directly involved in developing Sage or at least
> have them on sage-devel.
It's hard to get a lot of these people on sage-devel. Hello Mr.
Wolfram, we'd like some advice on developing a CAS. -- can you give us
some advice? That'll go over well. (... ok, who am I kidding, that'd
be a blast). Further, most experienced CAS developers have a project
to work on themselves -- otherwise we wouldn't know to ask them.
> A second is to identify these people and ask them what they think.
We do that a bit. But there's only so much input we want. One
problem in a big software project is that it's *WAY* too easy to just
spend all your time planning various tiny little details, and never
spend a minute writing code. I much prefer to do something
non-optimally than waste my time discussing it to the point that I
need to focus on another part of my life and the time I had for the
project is gone.
>A third is to examine other systems and see what
> kinds of advantages and disadvantages their designs entail (open
> source systems allows to read the code, too).
Well, it's rather late in the game to be looking at the underlying
design of Sage. It's built on python, it uses lots of cython, the
type system is pretty mature, and the last coercion rewrite took
almost a year and a handful of some the most committed sage developers
bleeding out the eyes and ears for nearly a week to review. Maybe a
few modules could use help, but Sage hasn't been in the "design" stage
for a long time.
> A fourth is to search the literature for anything relevant.
I think many of us here do exactly that.
> An with all of these except that
> first is that they delay development by adding steps beyond coding the
> first seemingly reasonable thing that comes to mind.
I've been witness to a fair amount of every single item you've
mentioned. But often times, the first seemingly reasonable thing that
comes to mind is exactly what gets implemented the first time. Then,
we improve where it makes sense.
> That is both good and bad.
Yup, and a natural balance will occur. I say, ain't broke, don't poke it.
> I thought to open this topic here because of a recent discussion on
> sage-flame, and I'm hesitant to do so since it has the potential to
> generate lots of talk to no benefit. On the other hand, it seems to me
> that if a good way can be found to benefit even a little more from
> past experience, that could have a significant impact on the quality
> of Sage. My impression is that asking the people with the relevant
> experience rarely happens, unless those people are already on sage-
> devel and choose to speak up, and doing more of that seems a low-
> effort way to benefit a little more from past experience.
Want to help? Get on trac and start reviewing tickets. This is how
we seek expert advice. See a problem, something done naively in the
source? Post a ticket, pop us an email, or fix it yourself. A number
of maintainers of packages Sage includes read sage-devel, and we hear
from them quite a bit. As William stated, your email indicates that
you don't really know much about Sage. Which is rather ironic -- as
an outsider, you've come in with the naive assumption that you know
best and you're going to tell these whippersnappers a thing or two
about what they're doing wrong -- in essence, done the first seemingly
Bjarke Hammersholt Roune
> mentioned. But often times, the first seemingly reasonable thing that
> comes to mind is exactly what gets implemented the first time. Then,
> we improve where it makes sense.
>
project.
> as an outsider, you've come in with the naive assumption that you know
> best and you're going to tell these whippersnappers a thing or two
>
I was careful to write my email in a way that does not imply that what
is going on now is other than best possible - or at least so I
thought.
William Stein
I think the problem is that Richard Fateman's repeated attacks on the
Sage project over the last four years have started to make many of us
perhaps overly defensive about anything that is even remotely similar.
I'm sure there is no harm done and I now hope that this will be a
productive and positive discussion.
-- William
rjf
on point.
Tom Boothby reveals himself once again as lacking in significant
comprehension,
in this case by thinking that the underlying design of Sage is ...
"it's built on Python ..."
Does this mean that prior experience building systems in Lisp, or C,
or Objective C, or C++ could not possibly be relevant to the Sage
design?
I would attribute this to some failure in communication, except that
Tom seems to be repeatedly voicing such ill-considered opinions that
there must be something other than
communication issues at work.
I learned only recently that the Sage type system and coercions
underwent a major and apparently uncomfortable revision. It would be
nice to know how the Sage type system compares to the systems in Axiom
and in Mupad, or for that matter, the systems that were more
experimental, e.g. Newspeak, Mode-Reduce, or even CLOS.
Is there a paper written on this topic? I would not be totally
surprised to find one of the following scenarios:
(a) Sage programmers did the first thing that came to mind, and after
finding it didn't work, redesigned it to look like one of the other
systems that pre-dated it. An example of "a year of programming can
save a day in the library."
(b) The Sage redesign falls somewhat short of one of the other
systems, and will either have to be redone yet again, or will simply
be less effective for expressing certain kinds of calculations for the
indefinite future.
........
On Sep 20, 6:51 am, William Stein <wst...@gmail.com> wrote:
> I think the problem is that Richard Fateman's repeated attacks on the
> Sage project over the last four years have started to make many of us
> perhaps overly defensive about anything that is even remotely similar.
> I'm sure there is no harm done and I now hope that this will be a
> productive and positive discussion.
just too subtle.
Cheers
RJF
William Stein
>
> I think that the comments from Bjarke Hammersholt Roune are directly
> on point.
> Tom Boothby reveals himself once again as lacking in significant
> comprehension,
> in this case by thinking that the underlying design of Sage is ...
> "it's built on Python ..."
>
> Does this mean that prior experience building systems in Lisp, or C,
> or Objective C, or C++ could not possibly be relevant to the Sage
> design?
> I would attribute this to some failure in communication, except that
> Tom seems to be repeatedly voicing such ill-considered opinions that
> there must be something other than
> communication issues at work.
>
> I learned only recently that the Sage type system and coercions
> underwent a major and apparently uncomfortable revision.
over a year ago, which is ancient history in Sage development time.
> It would be
> nice to know how the Sage type system compares to the systems in Axiom
> and in Mupad, or for that matter, the systems that were more
> experimental, e.g. Newspeak, Mode-Reduce, or even CLOS.
> Is there a paper written on this topic?
of the work on this topic mainly in their spare time. Robert's Ph.D.
at UW and David's at Harvard are both on topics that have nothing to
do with computer algebra, and for them to write a paper on this topic
would be too much of a distraction from real math.
> I would not be totally
> surprised to find one of the following scenarios:
>
> (a) Sage programmers did the first thing that came to mind, and after
> finding it didn't work, redesigned it to look like one of the other
> systems that pre-dated it. An example of "a year of programming can
> save a day in the library."
>
> (b) The Sage redesign falls somewhat short of one of the other
> systems, and will either have to be redone yet again, or will simply
> be less effective for expressing certain kinds of calculations for the
> indefinite future.
be redone.
So you're wrong, neither (a) nor (b) is the case.
> On Sep 20, 6:51 am, William Stein <wst...@gmail.com> wrote:
>> I think the problem is that Richard Fateman's repeated attacks on the
>> Sage project over the last four years have started to make many of us
>> perhaps overly defensive about anything that is even remotely similar.
>
> Wow! I have the power to make Tom rude to Bjarke!
>> I'm sure there is no harm done and I now hope that this will be a
>> productive and positive discussion.
>
> That's why you directed it to sage-flame? Sometimes your sarcasm is
> just too subtle.
discussion are actually nothing like yours. Your comments belong on
sage-flame.
>
> Cheers
>
> RJF
>
> >
>
--
William Stein
Associate Professor of Mathematics
University of Washington
http://wstein.org
Tom Boothby
> on point.
>
> Tom Boothby reveals himself once again as lacking in significant
> comprehension,
+5 points for statement of the obvious, Tom Boothby is often lacking
in significant comprehension, grad school is fraught with confusion --
math is hard!
-5 points for obvious lies, Tom Boothby does not reveal himself in
public, much less on the internet!
> in this case by thinking that the underlying design of Sage is ...
> "it's built on Python ..."
That was merely an example of an immutable design choice. Your powers
of miscomprehension in an email about poor communication astound me!
+2 points for ... what the heck?
> Does this mean that prior experience building systems in Lisp, or C,
> or Objective C, or C++ could not possibly be relevant to the Sage
> design?
-10 points for a baseless jump to conclusions! I said nothing of the sort!
-10 points for a failure to read the context! I think that lots of
people have a lot to offer us! Even you pipe up with something useful
from time to time!
> I would attribute this to some failure in communication, except that
> Tom seems to be repeatedly voicing such ill-considered opinions that
> there must be something other than
> communication issues at work.
-5 points for failure to recognize the character of the forum! If my
opinions were well-considered, I wouldn't post them to sage-flame!
> I learned only recently that the Sage type system and coercions
> underwent a major and apparently uncomfortable revision.
+1 point for not spending your time reading every single post on sage-devel!
> It would be
> nice to know how the Sage type system compares to the systems in Axiom
> and in Mupad, or for that matter, the systems that were more
> experimental, e.g. Newspeak, Mode-Reduce, or even CLOS.
+1 point for excellent idea! Compare them! I, for one, value
everything you contribute to sage-flame!
> Is there a paper written on this topic? I would not be totally
> surprised to find one of the following scenarios:
>
> (a) Sage programmers did the first thing that came to mind, and after
> finding it didn't work, redesigned it to look like one of the other
> systems that pre-dated it. An example of "a year of programming can
> save a day in the library."
>
> (b) The Sage redesign falls somewhat short of one of the other
> systems, and will either have to be redone yet again, or will simply
> be less effective for expressing certain kinds of calculations for the
> indefinite future.
>
>
> ........
>
>
>
>
> On Sep 20, 6:51 am, William Stein <wst...@gmail.com> wrote:
>
>
>> I think the problem is that Richard Fateman's repeated attacks on the
>> Sage project over the last four years have started to make many of us
>> perhaps overly defensive about anything that is even remotely similar.
>
> Wow! I have the power to make Tom rude to Bjarke!
I don't believe that I was being rude. At the very least, I posted to
sage-flame -- I certainly didn't mean to offend, and when I post here
I am overly harsh. It's fun. Oh -- I'm rude to you -- but I think
you (a) deserve it because you're rude to everybody anyway, and (b)
secretly enjoy it when people are rude to you.
rjf
On Sep 20, 9:44 pm, William Stein <wst...@gmail.com> wrote:
>
> > I think that the comments from Bjarke Hammersholt Roune are directly
> > on point.
>
> That explains why it was taken as such flame bait by everybody.
>
...
> > I learned only recently that the Sage type system and coercions
> > underwent a major and apparently uncomfortable revision.
>
> It was a lot of work. It wasn't "recent" though -- this all happened
> over a year ago, which is ancient history in Sage development time.
it recently.
>
> > It would be
> > nice to know how the Sage type system compares to the systems in Axiom
> > and in Mupad, or for that matter, the systems that were more
> > experimental, e.g. Newspeak, Mode-Reduce, or even CLOS.
>
> It's better?
with these systems.
>
> > Is there a paper written on this topic?
>
> No.
>Two Ph.D. students Robert Bradshaw and David Roe I think did most
> of the work on this topic mainly in their spare time.
> Robert's Ph.D.
> at UW and David's at Harvard are both on topics that have nothing to
> do with computer algebra, and for them to write a paper on this topic
> would be too much of a distraction from real math.
instead of incrementally modifying something that was pulled out of
someone's
posterior.
>
> > I would not be totally
> > surprised to find one of the following scenarios:
>
> > (a) Sage programmers did the first thing that came to mind, and after
> > finding it didn't work, redesigned it to look like one of the other
> > systems that pre-dated it. An example of "a year of programming can
> > save a day in the library."
>
> That's definitely *not* what happened.
>
>
>
> > (b) The Sage redesign falls somewhat short of one of the other
> > systems, and will either have to be redone yet again, or will simply
> > be less effective for expressing certain kinds of calculations for the
> > indefinite future.
>
> The Sage coercion model is very nice, and will definitely not have to
> be redone.
especially
about the future." (or something like that...)
>
> So you're wrong, neither (a) nor (b) is the case.
least coercions] is
"It's very nice".
>
> > On Sep 20, 6:51 am, William Stein <wst...@gmail.com> wrote:
> >> I think the problem is that Richard Fateman's repeated attacks on the
> >> Sage project over the last four years have started to make many of us
> >> perhaps overly defensive about anything that is even remotely similar.
>
> > Wow! I have the power to make Tom rude to Bjarke!
>
> Congratulations! Making Tom rude is quite an accomplishment. Not.
>
> >> I'm sure there is no harm done and I now hope that this will be a
> >> productive and positive discussion.
>
> > That's why you directed it to sage-flame? Sometimes your sarcasm is
> > just too subtle.
>
> I wrote the above after Bjarke explained that his goals in this
> discussion are actually nothing like yours. Your comments belong on
> sage-flame.
other (Bjarke) is actually nothing like that. So that does not
explain why you directed him to sage-flame :)
I have expressed my goals in terms of my comments about Sage and its
development.
William Stein
>> > nice to know how the Sage type system compares to the systems in Axiom
>> > and in Mupad, or for that matter, the systems that were more
>> > experimental, e.g. Newspeak, Mode-Reduce, or even CLOS.
>>
>> It's better?
>
> Spoken like a seeker for truth. I suspect that you are unfamiliar
> with these systems.
Are you familiar with them?
>>Two Ph.D. students Robert Bradshaw and David Roe I think did most
>> of the work on this topic mainly in their spare time.
>
> That's a qualification for designing a "type system" for Sage?
I made no such claim.
>> Robert's Ph.D.
>> at UW and David's at Harvard are both on topics that have nothing to
>> do with computer algebra, and for them to write a paper on this topic
>> would be too much of a distraction from real math.
>
> Yes, they would have to learn about type systems and computer algebra,
> instead of incrementally modifying something that was pulled out of
> someone's
> posterior.
They did no such thing.
>> >> I'm sure there is no harm done and I now hope that this will be a
>> >> productive and positive discussion.
>>
>> > That's why you directed it to sage-flame? Sometimes your sarcasm is
>> > just too subtle.
>>
>> I wrote the above after Bjarke explained that his goals in this
>> discussion are actually nothing like yours. Your comments belong on
>> sage-flame.
>
> Huh? So one of us (me) is guilty of providing flame-bait, and the
> other (Bjarke) is actually nothing like that. So that does not
> explain why you directed him to sage-flame :)
He carefully explained afterwards that he wanted to have a positive
and constructive discussion about specific ways to improve things.
> I have expressed my goals in terms of my comments about Sage and its
> development.
"Kids these days. Get off of my lawn."
William
Bill Hart
lists which offend people, I found the above discussion pretty
distasteful.
The entire concept of sage-flame is in my opinion unprofessional and
pretty rancid. I don't think you guys are doing yourselves any favours
by engaging in this.
I also felt that the knee-jerk reaction to Bjarke's original post was
completely off base.
To keep saying to people, "if you want to help out, deal with some
trac tickets or contribute some code" completely misses the point that
a number of people are making. Software engineering requires regular
introspection and revisiting of architectural and design issues. That
process is most definitely not accomplished by dealing with trac
tickets and contributing new code.
If you keep telling people like Bjarke to go away because they don't
know what they are talking about, and don't understand enough about
how Sage works, that is exactly what such people will do.
At the moment there's a group of people who are looking at Sage and:
1) Don't completely understand what Sage does, because it isn't
something you can just go look up
2) Would like to discuss the design and architectural components of
Sage
3) Have some very definite expertise to contribute
4) Are being encouraged to piss off
Bill.
rjf
On Sep 20, 10:03 pm, Tom Boothby <tomas.boot...@gmail.com> wrote:
>
> > I think that the comments from Bjarke Hammersholt Roune are directly
> > on point.
>
> > Tom Boothby reveals himself once again as lacking in significant
> > comprehension,
>
> +5 points for statement of the obvious, Tom Boothby is often lacking
> in significant comprehension, grad school is fraught with confusion --
> math is hard!
> -5 points for obvious lies, Tom Boothby does not reveal himself in
> public, much less on the internet!
>
> > in this case by thinking that the underlying design of Sage is ...
> > "it's built on Python ..."
>
> That was merely an example of an immutable design choice.
added ECL, and
have 100 packages in whatever languages they were written in.
> Your powers
> of miscomprehension in an email about poor communication astound me!
> +2 points for ... what the heck?
>
> > Does this mean that prior experience building systems in Lisp, or C,
> > or Objective C, or C++ could not possibly be relevant to the Sage
> > design?
>
> -10 points for a baseless jump to conclusions! I said nothing of the sort!
math students have, for the most part, avoided any significant level
of education (say a few graduate course in computer science --
languages, compilers, etc. Your ignorance is translated into
bluster.)
> -10 points for a failure to read the context! I think that lots of
> people have a lot to offer us!
Even you pipe up with something useful
> from time to time!
>
> > I would attribute this to some failure in communication, except that
> > Tom seems to be repeatedly voicing such ill-considered opinions that
> > there must be something other than
> > communication issues at work.
>
> > It would be
> > nice to know how the Sage type system compares to the systems in Axiom
> > and in Mupad, or for that matter, the systems that were more
> > experimental, e.g. Newspeak, Mode-Reduce, or even CLOS.
>
> +1 point for excellent idea! Compare them! I, for one, value
> everything you contribute to sage-flame!
system (or coercions or whatever changed), and the "final" perfect
solution as pronounced by William Stein.
The point that you seem to miss is that this is not unplowed
territory, but has been explored repeatedly, sometimes in great
details (even PhDs), and, so far as I can tell, you have ignored this
work, probably at your peril. Maybe you have a really good solution,
in which case you should be proud of it, and present it at a computer
science conference, instead of as PR at some SageDays N. If it has
even been presented there. But then you would have powerpoint slides
or something, but apparently you don't. Maybe the reason you are so
defensive about it is (a) you had nothing to do with it yourself, or
(b) you don't understand it, or both.
>
> I don't believe that I was being rude. At the very least, I posted to
> sage-flame -- I certainly didn't mean to offend,
> and when I post here
> I am overly harsh. It's fun.
have the knack.
Tom Boothby
>> sage-flame -- I certainly didn't mean to offend,
>
> Rudeness from someone I have no respect for is not offensive to me.
Ah, we're in agreement then.
>> and when I post here
>> I am overly harsh. It's fun.
>
> Rudeness without sarcasm is like a day without sunshine. You don't
> have the knack.
I'm from Seattle, you know. I prefer gray skies to protect my
sensitive eyes from the offensively bright sun. Besides, we do
sarcasm differently up here, too. They actually offer classes in
'passive aggressive driving' up here.
William Stein
>
> Whilst I may not be without fault when it comes to writing posts to
> lists which offend people, I found the above discussion pretty
> distasteful.
>
> The entire concept of sage-flame is in my opinion unprofessional and
> pretty rancid. I don't think you guys are doing yourselves any favours
> by engaging in this.
Please, please don't read or subscribe to sage-flame then! Nobody is
forcing you to read this list. It is important for people to have
an outlet like this, but it is certainly not important for you to read
it.
> To keep saying to people, "if you want to help out, deal with some
> trac tickets or contribute some code" completely misses the point that
> a number of people are making. Software engineering requires regular
> introspection and revisiting of architectural and design issues. That
> process is most definitely not accomplished by dealing with trac
> tickets and contributing new code.
>
> If you keep telling people like Bjarke to go away because they don't
> know what they are talking about, and don't understand enough about
> how Sage works, that is exactly what such people will do.
Nobody told Bjarke to go away.
> At the moment there's a group of people who are looking at Sage and:
>
> 1) Don't completely understand what Sage does, because it isn't
> something you can just go look up
Yes it is! That's the whole point. It's hard work, but a
sufficiently intelligent person can understand exactly what Sage does
in *any* case, because it is open source code. This is not the case
with Magma, Mathematics, Maple, Mupad, and Matlab, no matter how smart
you are.
> 2) Would like to discuss the design and architectural components of
> Sage
> 3) Have some very definite expertise to contribute
> 4) Are being encouraged to piss off
When somebody spends the time to actually genuinely *understand* Sage,
and then discusses the architecture, it sounds dramatically different
than when the people you're describing above try to discuss issues.
For example, the massive coercion model rewrite discussion 1-2 years
ago, along with the incremental improvements made to it by Nick
Thiery, were discussions by people with extremely extensive experience
with Sage and Python -- everybody involved had hundreds to thousands
of hours of experience with Python and the Sage library. Nick also
had tons of experience with MuPAD's type system (I think he actually
designed and implemented MuPAD's category system several years ago!)
Anyway, the point is that there are some people who think they have
something to offer, but haven't invested the time to understand much
about Sage, and who just sound silly to those who have. This
sometimes feels similar to people who think they have something to
offer toward a proof of the Riemann Hypothesis, say, since they know
some electrical engineering.
-- William
Robert Bradshaw
> On Sun, Sep 20, 2009 at 9:34 PM, rjf <fat...@gmail.com> wrote:
>>
>> I think that the comments from Bjarke Hammersholt Roune are directly
>> on point.
>
> That explains why it was taken as such flame bait by everybody.
I think it was more because of years worth of repeated (though
perhaps well-intentioned) accusations by one person that one part of
our system is being developed with insufficient regard to the past
has jaded many of us, especially as many of us do spend lots of time
reading research and looking at past and current software systems.
Or (c) All existing systems and designs were insufficient for our
needs, so we spent a lot of time coming up with something better.
The coercion model was the culmination of lots of discussion
involving many people, as well as looking at how many other systems
worked. I'm not claiming it's a novel breakthrough, but it works
well. The roughness is mostly in the fact that there are still
several parts of Sage that use the old-style APIs, and due to the
reasons mentioned above I personally haven't had time to work on this.
But perhaps you didn't think of option (c) because you reject the
notion that progress can be made, or at least that if progress can be
made, us folks who work on Sage are extremely unlikely to be the ones
making it.
- Robert
William Stein
>
>
>
> On Sep 20, 10:03 pm, Tom Boothby <tomas.boot...@gmail.com> wrote:
>> On Sun, Sep 20, 2009 at 9:34 PM, rjf <fate...@gmail.com> wrote:
>>
>> > I think that the comments from Bjarke Hammersholt Roune are directly
>> > on point.
>>
>> > Tom Boothby reveals himself once again as lacking in significant
>> > comprehension,
>>
>> +5 points for statement of the obvious, Tom Boothby is often lacking
>> in significant comprehension, grad school is fraught with confusion --
>> math is hard!
>> -5 points for obvious lies, Tom Boothby does not reveal himself in
>> public, much less on the internet!
>>
>> > in this case by thinking that the underlying design of Sage is ...
>> > "it's built on Python ..."
>>
>> That was merely an example of an immutable design choice.
>
> Except that you started with Python, and then added Cython,
Sage started with both Python and Cython originally (then called Pyrex).
>> Your powers
>> of miscomprehension in an email about poor communication astound me!
>> +2 points for ... what the heck?
>>
>> > Does this mean that prior experience building systems in Lisp, or C,
>> > or Objective C, or C++ could not possibly be relevant to the Sage
>> > design?
>>
>> -10 points for a baseless jump to conclusions! I said nothing of the sort!
>
> You did not answer the question. My guess is that you and your fellow
> math students have, for the most part, avoided any significant level
> of education (say a few graduate course in computer science --
> languages, compilers, etc. Your ignorance is translated into
> bluster.)
If computer scientists would design and implement good open source
math software, then we wouldn't have to do so.
>> +1 point for excellent idea! Compare them! I, for one, value
>> everything you contribute to sage-flame!
>
> Sure, you can start by writing a paper about the initial Sage type
> system (or coercions or whatever changed), and the "final" perfect
> solution as pronounced by William Stein.
I never said that the current solution is final or permanent. I only
said there won't be a total rewrite. There will certainly be small
refinements and improvements to the coercion model over time. But it
is too late in the lifecycle of Sage to be rewriting something at that
level, and nobody feels a need for it after using the new system for
over a year.
> The point that you seem to miss is that this is not unplowed
> territory,
> but has been explored repeatedly, sometimes in great
> details (even PhDs), and, so far as I can tell, you have ignored this
> work, probably at your peril.
You're the only one who keeps claiming that we "miss this". What do
you base that conclusion on? You just make stuff up. We're far too
lazy to *not* read papers on type systems and find out what other
programs do.
William
William Stein
<robe...@math.washington.edu> wrote:
>
> On Sep 20, 2009, at 9:44 PM, William Stein wrote:
>
>> On Sun, Sep 20, 2009 at 9:34 PM, rjf <fat...@gmail.com> wrote:
>>>
>>> I think that the comments from Bjarke Hammersholt Roune are directly
>>> on point.
>>
>> That explains why it was taken as such flame bait by everybody.
>
> I think it was more because of years worth of repeated (though
> perhaps well-intentioned) accusations by one person that one part of
> our system is being developed with insufficient regard to the past
> has jaded many of us, especially as many of us do spend lots of time
> reading research and looking at past and current software systems.
could find about every computer algebra system out there. And when I
ran out of things to read I had long conversations with the people
that wrote them. I think I'm not at all unusual among Sage
developers in this regards.
-- William
Harald Schilly
On Sep 21, 4:30 pm, rjf <fate...@gmail.com> wrote:
> Except that you started with Python, and then added Cython, and then
> added ECL, and
> have 100 packages in whatever languages they were written in.
so what do you expect? Everything only lisp?!? This is getting so so
so booring ... Guess what, Sage would not need ECL if there is one
package less on the boat. And the other languages aren't that many, in
fact, they are written in very popular ones and Sage's approach is to
use a two-layer model: a higher level to hold things togehter, and a
lower level with "local" responsibilites, in a non interpreted
language.
H
rjf
On Sep 21, 10:35 am, William Stein <wst...@gmail.com> wrote:
> It's hard work, but a
> sufficiently intelligent person can understand exactly what Sage does
> in *any* case, because it is open source code. This is not the case
> with Magma, Mathematics, Maple, Mupad, and Matlab, no matter how smart
> you are.
designed a type system
without studying Axiom, you have left a few decades of work
unexamined. It is not too late.
As for the various efforts to duplicate Maxima so you won't have to
call it, I think it would
be helpful, occasionally, to have someone who has "sufficient"
intelligence to read the code.
Even if it is in Lisp.
>
> > 2) Would like to discuss the design and architectural components of
> > Sage
> > 3) Have some very definite expertise to contribute
> > 4) Are being encouraged to piss off
>
> When somebody spends the time to actually genuinely *understand* Sage,
> and then discusses the architecture, it sounds dramatically different
> than when the people you're describing above try to discuss issues.
people.
It is implausible that anyone would need to know all of that,
especially since
at least one module that I am somewhat familiar with is fairly
complex.
You seem to think that describing something like a type system is too
hard and has to be understood by reading source code. I find that hard
to believe.
> For example, the massive coercion model rewrite discussion 1-2 years
> ago, along with the incremental improvements made to it by Nick
> Thiery, were discussions by people with extremely extensive experience
> with Sage and Python --
Maxima and therefore am an expert who can write the package for
computation of Kloosterman sums and Fourier coefficients of cusp
forms, you might say, Whoa. What makes you an expert on THAT?
Now consider the reverse proposal. You have all these math-y types
who know whatever.
You think that therefore they magically "know" whatever is relevant in
computer science.
So you know how to write simple programs. Does that mean you know
about software, type systems, compiling, language design, etc?
I know how to add, subtract, multiply. I even know how to compute
determinants. Does that make me just as good as an expert on Plucker
forms??
> everybody involved had hundreds to thousands
> of hours of experience with Python and the Sage library.
hours with Scheme.
Collectively over a year they have a hundred thousand hours. Does
this make them experts? Well, by and large, no. But some may have far
better instincts than your volunteers.
The ones who go on to take graduate courses in (say) the design or
implementation of programming languages, are exposed to a wide range
of issues and solutions to problems.
> Nick also
> had tons of experience with MuPAD's type system (I think he actually
> designed and implemented MuPAD's category system several years ago!)
compare!
Instead all we get here is bluster from you.
>
> Anyway, the point is that there are some people who think they have
> something to offer, but haven't invested the time to understand much
> about Sage, and who just sound silly to those who have.
much of their time on public relations, silly projects, and attempts
to
deflect (even constructive) comments.
> This
> sometimes feels similar to people who think they have something to
> offer toward a proof of the Riemann Hypothesis, say, since they know
> some electrical engineering.
to
offer towards the design of computer algebra systems, say, since they
know
arithmetic and one programming language.
RJf
>
rjf
On Sep 21, 10:40 am, Robert Bradshaw <rober...@math.washington.edu>
wrote:
>
> Or (c) All existing systems and designs were insufficient for our
> needs, so we spent a lot of time coming up with something better.
>
> The coercion model was the culmination of lots of discussion
> involving many people, as well as looking at how many other systems
> worked. I'm not claiming it's a novel breakthrough, but it works
> well. The roughness is mostly in the fact that there are still
> several parts of Sage that use the old-style APIs, and due to the
> reasons mentioned above I personally haven't had time to work on this.
> But perhaps you didn't think of option (c) because you reject the
> notion that progress can be made, or at least that if progress can be
> made, us folks who work on Sage are extremely unlikely to be the ones
> making it.
suppose
it is pointless to discuss how it compares to Axiom or Mupad, each of
which
has been described.
>
> - Robert
rjf
On Sep 21, 10:45 am, William Stein <wst...@gmail.com> wrote:
>
> You're the only one who keeps claiming that we "miss this". What do
> you base that conclusion on? You just make stuff up. We're far too
> lazy to *not* read papers on type systems and find out what other
> programs do.
what other
people had done, and failed to do?
I don't think you are lazy. Arrogant, maybe.
William Stein
>
>
>
> On Sep 21, 10:35 am, William Stein <wst...@gmail.com> wrote:
>> It's hard work, but a
>> sufficiently intelligent person can understand exactly what Sage does
>> in *any* case, because it is open source code. This is not the case
>> with Magma, Mathematics, Maple, Mupad, and Matlab, no matter how smart
>> you are.
>
> Well, you significantly left out Axiom and Maxima.
I made the statement "a sufficiently intelligent person can understand
what Sage does because it is open source but not ...". I left out
Axiom and Maxima, since they are also open source.
> As for the various efforts to duplicate Maxima so you won't have to
> call it, I think it would
> be helpful, occasionally, to have someone who has "sufficient"
> intelligence to read the code.
We do look at the Maxima code.
> Even if it is in Lisp.
>> > 2) Would like to discuss the design and architectural components of
>> > Sage
>> > 3) Have some very definite expertise to contribute
>> > 4) Are being encouraged to piss off
>>
>> When somebody spends the time to actually genuinely *understand* Sage,
>> and then discusses the architecture, it sounds dramatically different
>> than when the people you're describing above try to discuss issues.
>
> I understand that Sage has 100 (or is it 83?) modules written by other
> people.
> It is implausible that anyone would need to know all of that,
> especially since
> at least one module that I am somewhat familiar with is fairly
> complex.
>
> You seem to think that describing something like a type system is too
> hard and has to be understood by reading source code. I find that hard
> to believe.
Unfortunately, it is impossible to really understand the
*implementation* of Sage's "type system" (whatever that means) without
reading the code. This is unfortunate, and it's not a point of
pride. It's just the way it is right now.
>> For example, the massive coercion model rewrite discussion 1-2 years
>> ago, along with the incremental improvements made to it by Nick
>> Thiery, were discussions by people with extremely extensive experience
>> with Sage and Python --
>
> Well, if I said I had extremely extensive experience with Python and
> Maxima and therefore am an expert who can write the package for
> computation of Kloosterman sums and Fourier coefficients of cusp
> forms, you might say, Whoa. What makes you an expert on THAT?
I would say, "Implement it and post a patch."
> Now consider the reverse proposal. You have all these math-y types
> who know whatever.
> You think that therefore they magically "know" whatever is relevant in
> computer science.
No I don't think that.
> So you know how to write simple programs. Does that mean you know
> about software, type systems, compiling, language design, etc?
> I know how to add, subtract, multiply. I even know how to compute
> determinants. Does that make me just as good as an expert on Plucker
> forms??
"Those who can -- do. Those who can't -- teach." I don't care if
somebody is a licensed and certified expert or not, what matters is
what they are able to do.
>> everybody involved had hundreds to thousands
>> of hours of experience with Python and the Sage library.
>
> You know, each member of my freshman/sophomore class has a hundred
> hours with Scheme.
> Collectively over a year they have a hundred thousand hours. Does
> this make them experts? Well, by and large, no. But some may have far
> better instincts than your volunteers.
> The ones who go on to take graduate courses in (say) the design or
> implementation of programming languages, are exposed to a wide range
> of issues and solutions to problems.
Good for them!
>
>
>> Nick also
>> had tons of experience with MuPAD's type system (I think he actually
>> designed and implemented MuPAD's category system several years ago!)
>
> Well, then maybe Nick would write a short paper describing how they
> compare!
Maybe he will. That's up to him. H might not get academic credit for
it. In a mathematics department faculty evaluation, a paper on "type
systems in computer algebra" might not count for much. He is not a
computer algebra researcher, but a mathematician, who writes software
because all software that is out there already is not capable of
solving the problems he confronts in his research. I suspect
publishing computer science papers is far from his mind.
>> Anyway, the point is that there are some people who think they have
>> something to offer, but haven't invested the time to understand much
>> about Sage, and who just sound silly to those who have.
>
> That may be because the ones who are boosting Sage are spending so
> much of their time on public relations, silly projects, and attempts
> to deflect (even constructive) comments.
So perhaps if we stop spending time on public relations, silly
projects, and attempts to
deflect comments, then major architectural suggestions from people who
are not familiar with Sage will somehow start making sense?
>> This
>> sometimes feels similar to people who think they have something to
>> offer toward a proof of the Riemann Hypothesis, say, since they know
>> some electrical engineering.
>
> This definitely feels similar to people who think they have something
> to
> offer towards the design of computer algebra systems, say, since they
> know
> arithmetic and one programming language.
I do not claim to have something academic to say about the design of
computer algebra systems. If so, I would be trying to publish
papers on them, speak at conference about them, etc. I just
want mathematics software that does what I need now for purely
practical purposes. I'm not a computer algebra researcher. I'm a
number theorist and by necessity a software engineer. I definitely
don't claim to be a computer scientist.
Computer science is a fine discipline; I have nothing against it.
-- William
William Stein
It wasn't really so much of a rewrite as iterative development. We
first used the Python coercion model. Then (mainly) David Harvey
designed a more sophisticated one at Sage Days 2, which I implemented
and rolled out. There was a very painful attempt by certain people
to do a full rewrite, but that ended in "disaster", was tossed, and
then finally last year Robert Bradshaw did another *iterative*
redesign and implementation, which is what we use now.
> I don't think you are lazy. Arrogant, maybe.
"Ad hominem abusive (also called argumentum ad personam) usually and
most notoriously involves insulting or belittling one's opponent, but
can also involve pointing out factual but ostensible character flaws
or actions which are irrelevant to the opponent's argument."
-- Wikipedia
rjf
On Sep 21, 12:45 pm, Harald Schilly <harald.schi...@gmail.com> wrote:
> On Sep 21, 4:30 pm, rjf <fate...@gmail.com> wrote:
>
> > Except that you started with Python, and then added Cython, and then
> > added ECL, and
> > have 100 packages in whatever languages they were written in.
>
> How Sage has been started? The idea is to combine existing software,
> so what do you expect? Everything only lisp?!?
I did hope that the criticisms of Lisp as an implementation language
for computer
algebra would have some basis in fact.
> so booring ...
>Guess what, Sage would not need ECL if there is one
> package less on the boat.
Someone could start advertising Maxima as an open-source viable
alternative to Mathematica, Maple, Matlab.. And if someone said "what
about Sage?" we could say, oh there is a newsgroup maxima-flame, ask
there.
:)
>
Tom Boothby
> Someone could start advertising Maxima as an open-source viable
> alternative to Mathematica, Maple, Matlab.. And if someone said "what
> about Sage?" we could say, oh there is a newsgroup maxima-flame, ask
> there.
> :)
Oooh ooh can you?!? It'd be awesome of you to go and piss on your own
floor for a while. We wouldn't have sage-flame if it wasn't for you,
y'know.
Robert Bradshaw
I subscribe to sage-flame for the discussions that are too flame-
baitish for sage-devel, but still very interesting, not for outright
attacks.
- Robert
Tom Boothby
> baitish for sage-devel, but still very interesting, not for outright
> attacks.
Hmm. I rather enjoy the flaming, personally. To each his own, I
guess. There was a day when the only flamebait was licensing
discussions -- those were the days -- I can really get behind a
serious discussion about licensing matters, and I think it's an
important thing for any open-source developer to think about. Hence,
I think those sorts of topics should be discussed in the open on
sage-devel. But, some people just can't be civil, so was born
sage-flame -- and some of us don't mind sinking to that level once and
a while for the fun of it.
William Stein
It is worth discussing what the point of "sage-flame" is. First, here
is the Wikipedia definition of "flame":
"Flaming is a hostile and insulting interaction between Internet users."
It then goes on: "An Internet user typically generates a flame
response to other posts or users posting on a site, and such a
response is usually not constructive, does not clarify a discussion,
and does not persuade others. Sometimes, flamers attempt to assert
their authority, or establish a position of superiority over other
users. Other times, a flamer is simply an individual who believes he
or she carries the only valid opinion. This leads him or her to
personally attack those who disagree. In some cases, flamers wish to
upset and offend other members of the forum, in which case they can be
called "trolls". Most often however, flamers are angry or insulting
messages transmitted by people who have strong feelings about a
subject."
I created sage-flame to be an outlet for such lame and ugly behavior,
mainly because such behavior absolutely does not belong on sage-devel
or any of the other sage mailing lists. However, this behavior has
come up repeatedly over the years, and people being what they are, it
will come up again. Instead of wasting time policing such discussions
on sage-devel, we have this list where anything goes.
-- William
Ondrej Certik
I think anything should be allowed on sage-flame.
Ondrej
Robert Bradshaw
Yeah, I realize that. I wasn't even going to subscribe at first...but
then actual non-insulting interesting discussion started to happen. I
guess I started to expect a place where controversial topics like
licenses and languages could be discussed in a sane manner most of
the time.
I see the value in a list where anything goes, and what better name
than sage-flame.
- Robert
Ondrej Certik
That's right, I see this the main goal too.
>
> I see the value in a list where anything goes, and what better name
> than sage-flame.
So I think there should not be a priori set rules about behavior, but
we (e.g. me as Ondrej, you as Robert, etc.) should not tolerate
unnaceptable behavior, so that we can still discuss things.
Ondrej
William Stein
Maybe that could happen in yet another list :-)
Maybe sage-design... which would be civil and sane discussions of the
basic design decisions in Sage. Making and managing Google groups is
really easy.
William
Ondrej Certik
What about licensing? I think sage-flame is a good list for all of
this. Let's leave it like this.
Ondrej
rjf
On Sep 21, 9:14 pm, Tom Boothby <tomas.boot...@gmail.com> wrote:
Maxima "off the boat."
Tom Boothby
This would be a lot less fun if I read everything you wrote.
Bill Hart
On 21 Sep, 18:35, William Stein <wst...@gmail.com> wrote:
>
> > Whilst I may not be without fault when it comes to writing posts to
> > lists which offend people, I found the above discussion pretty
> > distasteful.
>
> > The entire concept of sage-flame is in my opinion unprofessional and
> > pretty rancid. I don't think you guys are doing yourselves any favours
> > by engaging in this.
>
> Please, please don't read or subscribe to sage-flame then! Nobody is
> forcing you to read this list. It is important for people to have
> an outlet like this, but it is certainly not important for you to read
> it.
once!
>
> > To keep saying to people, "if you want to help out, deal with some
> > trac tickets or contribute some code" completely misses the point that
> > a number of people are making. Software engineering requires regular
> > introspection and revisiting of architectural and design issues. That
> > process is most definitely not accomplished by dealing with trac
> > tickets and contributing new code.
>
> > If you keep telling people like Bjarke to go away because they don't
> > know what they are talking about, and don't understand enough about
> > how Sage works, that is exactly what such people will do.
>
> Nobody told Bjarke to go away.
the treatment of this regular, enthusiastic Sage contributor to be
very distasteful.
>
> > At the moment there's a group of people who are looking at Sage and:
>
> > 1) Don't completely understand what Sage does, because it isn't
> > something you can just go look up
>
> Yes it is! That's the whole point. It's hard work, but a
> sufficiently intelligent person can understand exactly what Sage does
> in *any* case, because it is open source code. This is not the case
> with Magma, Mathematics, Maple, Mupad, and Matlab, no matter how smart
> you are.
Sage operates, would like to discuss the design or would like some
extra guidance on the code you have written, your answer is they
should go away and read the code and that they haven't taken the time
to understand it properly.
You know full well there are people who would like to understand
various components of Sage, who have tried to read your code to
understand what it does and still need guidance.
Imagine if I did that with FLINT or MPIR every time someone asked how
it worked. If I said, "oh you didn't take the time to understand the
MPIR system. Go and get yourself a clue". Imagine how arrogant people
would think I was. Or if I just assumed it was by definition the best
thing that could be done because it was the MPIR way and someone spent
a long time thinking it through so it must be good.
>
> > 2) Would like to discuss the design and architectural components of
> > Sage
> > 3) Have some very definite expertise to contribute
> > 4) Are being encouraged to piss off
>
> When somebody spends the time to actually genuinely *understand* Sage,
> and then discusses the architecture, it sounds dramatically different
> than when the people you're describing above try to discuss issues.
> For example, the massive coercion model rewrite discussion 1-2 years
> ago, along with the incremental improvements made to it by Nick
> Thiery, were discussions by people with extremely extensive experience
> with Sage and Python -- everybody involved had hundreds to thousands
> of hours of experience with Python and the Sage library. Nick also
> had tons of experience with MuPAD's type system (I think he actually
> designed and implemented MuPAD's category system several years ago!)
>
> Anyway, the point is that there are some people who think they have
> something to offer, but haven't invested the time to understand much
> about Sage, and who just sound silly to those who have. This
> sometimes feels similar to people who think they have something to
> offer toward a proof of the Riemann Hypothesis, say, since they know
> some electrical engineering.
>
is that without hundreds to thousands of hours experience with Sage
and Python, they are not qualified to discuss Sage from a software
engineering perspective. And even sometimes when people come along who
do have hundreds of hours of experience, if they ask a question about
the design of Sage they are treated like imbeciles because they
weren't involved in the actual design process and therefore can't
possibly have the required experience.
Imagine if mathematics were like that. Imagine if someone with
experience in ergodic theory were treated like an imbecile by someone
who developed the theory of stacks because they weren't involved in
the formative development of that subject, and if they asked, "can you
explain to me what it is you've done", or "do you have a survey on
this subject", or "how does what you've done compare to such and such"
they were told, "you aren't qualified to discuss this because you
didn't take the time to understand my annals papers, or any of my
copious handwritten research notes which I've scanned and put on the
web". Oh wait, mathematics is like that...
The other thing that has really pissed me off here is you have taken a
discussion by a regular Sage developer and contributor and made it
about Richard Fateman. Pretty much exactly the same thing happened
when my discussion got unceremoniously copied to sage-flame and
publicly insulted. If your intention is to alienate Sage contributors,
that's a pretty good strategy. It's also unnecessarily rude to
Fateman. It's possible to answer Fateman without resorting to public
insult (ok, I may have done this myself once or twice, so I am not
without blame here, but I have written to Fateman to let him know I
have taken the time to think through some of the points that he made,
read up on Lisp and Maxima and discovered some of his points are not
without some modicum of merit - though I have been clear that I do not
consider Maxima to hold any particular appeal to me as a number
theorist and that Lisp is syntactically unnatural and sometimes
inconsistent).
Regardless of what I may think of Fateman's opinions, I think the
practice of publicly insulting regular Sage contributors by moving
their discussions to sage-flame to be odious and pretentious. I object
in the strongest possible terms.
I also have some more specific objections, primarily to do with the
way Fateman himself is being answered. I think there is a little bit
of a pretence that Sage does have the answers to all the objections of
Fateman. I think in some cases that just isn't true. In some cases it
might be better to simply admit that these are problems Sage has not
attempted to deal with and in other cases simply admit that Sage has
grown organically, often without planning (in the sense that Fateman
means it). Numerous times code has been removed from Sage because it
wasn't up to snuff. In some cases code has been added to Sage, even
when it was known to be rubbish, simply because it represented work
contributed by someone who put a lot of (perhaps misguided) effort in.
You (William) have numerous times said that basically you prefer
people to learn by trying things out than by having experts sit around
and thrash out a highly engineered solution. That is, by and large,
how Sage has been constructed. You have to understand that when that
is, by and large, the way Sage has historically developed, it is a
little bit too much to expect people to believe that the type system
in Sage has been any different. It might well have been different. But
you've got to accept that there is going to be some scepticism when
you claim that in fact the type system is highly thought out and
carefully designed with due regard to all the previous efforts along
those lines. Extraordinary claims require extraordinary evidence, and
on one point I will explicitly, publicly agree with Fateman. The best
way to describe this new, highly innovative, well thought out type
system in Sage is to write a (survey) paper on the subject and publish
it.
If people don't have time to write papers on it because it is too far
removed from the area they work in, then perhaps, just perhaps, the
people who designed that system are not the experts you claim them to
be. Or at the very least, you have to accept that people will believe
that to be the case. It just isn't credible to say to an expert on
type systems, "oh, you don't have the right to criticise our type
system because you don't have hundreds to thousands of hours
experience developing Sage and Python, and if you took the time to
read the source code you'd figure out it is actually a really
brilliant, well thought out system designed by experts". The only way
such a claim is going to be credible is if a technical paper on the
subject exists that you can point such a person to. Maybe, just maybe,
Fateman knows enough of Nick Thiery's work on type systems to adjudge
that the system used by Sage is likely to be technically very sound.
It will be interesting to see how Fateman responds to that. Even
still, it is a bit of a stretch for me to believe that Nick is such an
expert, but will not write a paper on the subject describing the
innovations because it is too far out of his field. You can't have it
both ways. I don't claim Nick is not such an expert. I just claim that
pretending that it is Fateman's lack of an understanding of the Sage
source code or his lack of hundreds to thousands of hours developing
Sage Python that is the problem here, is a bit far fetched.
Bill.
Bill Hart
take it from someone older than you and perhaps with more experience.
You are being sorely misled here. Insulting professionals publicly is
an extremely distasteful and ill-advised thing to do, regardless of
where it is done.
Everything you write on these lists is being recorded in a very public
and permanent way. And everything you do, write or say here has the
potential to affect your career.
I don't know who is leading you to believe that insulting people is
funny, clever or the right thing to do, but as a friend and older
colleague I would counsel you to think carefully before doing it. I
hope I have not been one of the people who has misled you in this
regard.
Bill.
Bill Hart
and enthusiastic sage developers to sage-flame and insulting them?
What about insulting professionals publicly and encouraging others to
do so?
What about ad hominem attacks?
I just can't believe what I am reading here! I held all of you in much
higher esteem.
Let me give you all a hint. Someone was deeply insulted here at a
particularly low point in their life. What you are doing here is
deeply destructive. You might all think it is funny, but I can tell
you for a fact that you are really upsetting people.
Bill.
On 22 Sep, 05:56, Ondrej Certik <ond...@certik.cz> wrote:
> On Mon, Sep 21, 2009 at 9:47 PM, William Stein <wst...@gmail.com> wrote:
>
> > On Mon, Sep 21, 2009 at 9:33 PM, Robert Bradshaw
> > <rober...@math.washington.edu> wrote:
>
> >> On Sep 21, 2009, at 9:14 PM, Tom Boothby wrote:
>
William Stein
>
>
>
> On 21 Sep, 18:35, William Stein <wst...@gmail.com> wrote:
>> On Mon, Sep 21, 2009 at 3:54 AM, Bill Hart <goodwillh...@googlemail.com> wrote:
>>
>> > Whilst I may not be without fault when it comes to writing posts to
>> > lists which offend people, I found the above discussion pretty
>> > distasteful.
>>
>> > The entire concept of sage-flame is in my opinion unprofessional and
>> > pretty rancid. I don't think you guys are doing yourselves any favours
>> > by engaging in this.
>>
>> Please, please don't read or subscribe to sage-flame then! Nobody is
>> forcing you to read this list. It is important for people to have
>> an outlet like this, but it is certainly not important for you to read
>> it.
>
> Umm, hello!! The discussion got moved here from sage-devel - more than
> once!
You do not have to read this list, even if a discussion is moved to it.
>> > 1) Don't completely understand what Sage does, because it isn't
>> > something you can just go look up
>>
>> Yes it is! That's the whole point. It's hard work, but a
>> sufficiently intelligent person can understand exactly what Sage does
>> in *any* case, because it is open source code. This is not the case
>> with Magma, Mathematics, Maple, Mupad, and Matlab, no matter how smart
>> you are.
>
> So, your premise here is that when someone asks questions about how
> Sage operates, would like to discuss the design or would like some
> extra guidance on the code you have written, your answer is they
> should go away and read the code and that they haven't taken the time
> to understand it properly.
No. Fateman claimed Sage is not "something you can just go look up."
I pointed out that this is false.
NO. I merely remarked that sometimes when people go on and on about
something that they don't have significant experience with then they
can sound silly to people with real experience.
> Pretty much exactly the same thing happened
> when my discussion got unceremoniously copied to sage-flame and
> publicly insulted.
You specifically told me in that email that I could cc sage-flame. I like to
keep discussions onlist whenever possible.
> Regardless of what I may think of Fateman's opinions, I think the
> practice of publicly insulting regular Sage contributors by moving
> their discussions to sage-flame to be odious and pretentious. I object
> in the strongest possible terms.
So moving any discussion to sage-flame is a "public insult"?
I just want a viable free open source alternative to Magma,
Mathematica, Maple, and Matlab. I could care less about publishing
a CS paper to establish some "claim" of doing something cool.
> If people don't have time to write papers on it because it is too far
> removed from the area they work in, then perhaps, just perhaps, the
> people who designed that system are not the experts you claim them to
> be.
I don't care if somebody is an expert or not. I do care if they can
create software that helps make a viable alternative to the MA's.
> Or at the very least, you have to accept that people will believe
> that to be the case. It just isn't credible to say to an expert on
> type systems, "oh, you don't have the right to criticise our type
> system because you don't have hundreds to thousands of hours
> experience developing Sage and Python, and if you took the time to
> read the source code you'd figure out it is actually a really
> brilliant, well thought out system designed by experts".
So far nobody has criticized Sage's type system. In fact the actual
system hasn't even come up in any of this thread. What is being
criticized evidently is the lack of a paper about it. I very honestly
explained why there isn't such a paper, repeatedly. I'm not
claiming it is good or bad that there isn't a paper. It's just a fact
that there isn't one.
William
William Stein
>
> That includes moving discussions of well-meaning, carefully considered
> and enthusiastic sage developers to sage-flame and insulting them?
>
> What about insulting professionals publicly and encouraging others to
> do so?
>
> What about ad hominem attacks?
>
> I just can't believe what I am reading here! I held all of you in much
> higher esteem.
>
> Let me give you all a hint. Someone was deeply insulted here at a
> particularly low point in their life. What you are doing here is
> deeply destructive. You might all think it is funny, but I can tell
> you for a fact that you are really upsetting people.
>
> Bill.
In order to make you happy, I'm changing the settings of the
sage-flame mailing list. It will henceforth be a private list with a
private archive not visible to non-members. New
members must request an invitation to join.
This should answer your issues with:
(1) your perception of public insults,
(2) your being made to feel uncomfortable by this list (I've
unsubscribed you), and
(3) your concerns about Tom Boothby's future.
I think there are people who are able to engage in heated discussions,
arguments and flames in a group specifically created for that purpose.
However, your concern that this is should not be done in public are
well founded, so out of respect for that I've closed sage-flame to the
public.
-- William
rjf
On Sep 22, 4:17 am, William Stein <wst...@gmail.com> wrote:
> However, your concern that this is should not be done in public are
> well founded, so out of respect for that I've closed sage-flame to the
> public.
>
> -- William
note to Tom.)
I'm not sure what to do here --
Closing the list for reading is, I think, unfortunate. People will
see pointers to it from sage-devel or perhaps other places. Some
people have written commentary (or screeds...)
specifically with the assumption that their comments will be available
to interested parties.
So this move, to me,
is like "thanks for sharing your comments. they have been archived
on /dev/null."
I don't know what the options are for messages, but if the concern is
for Tom, can Tom (or someone) delete messages that are deemed by Tom
(or others?) ill advised. I don't propose to scan for embedded
quotes, since that would be kind of tricky, unless you rewrite other
peoples' notes.
RJF
rjf
written by Tom.
I suppose Tom could request others to delete or edit their messages.
Years ago, when people got their first email accounts on
arrival at college, some students chose login names that eventually
would make them uncomfortable, e.g. 4 years later on their
resume. ...
"Seeking well-paying job in analytical financial firm, for more
information
contact stewe...@berkeley.edu..."
The university changed its policy to allow students to rename their
email accounts. Once.
Now, of course, stewed dude would post a photo of himself nude, drunk,
and perhaps
vomiting, on facebook.
William Stein
>
> Oh, to clarify -- I mean Tom should be free to delete messages
> written by Tom.
> I suppose Tom could request others to delete or edit their messages.
Me -- list closed
You -- list opened
So... Tom Boothby, can you make the tie breaking vote?
William
>
> Years ago, when people got their first email accounts on
> arrival at college, some students chose login names that eventually
> would make them uncomfortable, e.g. 4 years later on their
> resume. ...
> "Seeking well-paying job in analytical financial firm, for more
> information
> contact stewe...@berkeley.edu..."
>
> The university changed its policy to allow students to rename their
> email accounts. Once.
> Now, of course, stewed dude would post a photo of himself nude, drunk,
> and perhaps
> vomiting, on facebook.
>
>
>
> >
>
William Stein
Associate Professor of Mathematics
University of Washington
http://wstein.org
Tom Boothby
line there. If he took offense, he didn't say so, and I don't feel
any remorse there. The following catfight with RJF had nothing to do
with Bjarke, and I have no remorse for what happened there.
Given the nature of this board, and the fact that people get their
panties in a tangle over what happens here -- even though we call it
sage-flame -- we should close it. Nothing of value should happen
here, so nothing of value will be lost.
--tom
Robert Bradshaw
I completely agree with you here. As stated, the purpose of sage-
flame was to be able to say "that kind of discussion doesn't belong
here, please move it to sage-flame" which, in retrospect, would be
meaningless if we policed sage-flame as well (though hopefully some
degree of self-policing should always go on). However, what started
out as a tool for diverting and suppressing this kind of behavior has
turned into a tool for encouraging it, and as you mentioned, non-
flame discussions have repeatedly been moved here. Sounds like we
need a new list.
- Robert
William Stein
>
> Re-reading what I wrote to Bjarke, I don't believe that I was out of
> line there. If he took offense, he didn't say so, and I don't feel
> any remorse there. The following catfight with RJF had nothing to do
> with Bjarke, and I have no remorse for what happened there.
>
> Given the nature of this board, and the fact that people get their
> panties in a tangle over what happens here -- even though we call it
> sage-flame -- we should close it. Nothing of value should happen
> here, so nothing of value will be lost.
has to approve new members.
William
Robert Bradshaw
> On Sep 22, 4:17 am, William Stein <wst...@gmail.com> wrote:
>
>> However, your concern that this is should not be done in public are
>> well founded, so out of respect for that I've closed sage-flame to
>> the
>> public.
>>
>> -- William
>
> I agree with Bill's recent posts 35,36,37 in my listing. (including
> note to Tom.)
> I'm not sure what to do here --
>
> Closing the list for reading is, I think, unfortunate. People will
> see pointers to it from sage-devel or perhaps other places. Some
> people have written commentary (or screeds...)
> specifically with the assumption that their comments will be available
> to interested parties.
>
> So this move, to me,
> is like "thanks for sharing your comments. they have been archived
> on /dev/null."
I agree this is unfortunate. sage-flame has been the forum for both
heated debate and ad hominem attackes. Perhaps some people don't make
the distinction, but in my mind the former are sometimes interesting
and worth preserving, while the latter deserved to be filed in /dev/
null (perhaps before they even get read :). Neither belongs on sage-
devel (even if it's just to try to preserve the signal-noise ratio
for those not interested).
- Robert
William Stein
It's not exactly /dev/null. Any member can read the full archives.
William
Harald Schilly
> Everything you write on these lists is being recorded in a very public
> and permanent way. And everything you do, write or say here has the
> potential to affect your career.
work for anyone who judges me by reading what i have written years ago
on a *flame mailing list. I'm sure things wil change and it will be a
sign of concern, if there is nothing about you in the internet. I even
guess that a future president of the usa already told us on facebook
about his last party!
H
rjf
On Sep 22, 12:38 pm, Harald Schilly <harald.schi...@gmail.com> wrote:
> On Sep 22, 12:43 pm, Bill Hart <goodwillh...@googlemail.com> wrote:
>
> > Everything you write on these lists is being recorded in a very public
> > and permanent way. And everything you do, write or say here has the
> > potential to affect your career.
>
> Sorry, but, my personal opinion on that is this: i would not want to
> work for anyone who judges me by reading what i have written years ago
> on a *flame mailing list.
future?
rjf
On Sep 22, 11:21 am, William Stein <wst...@gmail.com> wrote:
> It's not exactly /dev/null. Any member can read the full archives.
>
RJF
Tom Boothby
A: A duck's got feathers.
William Stein
>
> Q: What's the difference between a duck?
> A: A duck's got feathers.
and useful.
Elephants with feathers are a curiosity. Is SAGE an elephant
with feathers?
And the future is pretty grim."
RJF -- 2005, http://sci.tech-archive.net/Archive/sci.math.symbolic/2005-12/msg00096.html
>
> On Tue, Sep 22, 2009 at 5:26 PM, rjf <fat...@gmail.com> wrote:
>>
>>
>>
>> On Sep 22, 11:21 am, William Stein <wst...@gmail.com> wrote:
>>
>>> It's not exactly /dev/null. Any member can read the full archives.
>>>
>>
>> I refer you to my earlier post, Welcome to the Kailee.
>> RJF
>>
>> >
>>
>
> >
>
rjf
On Sep 22, 10:16 pm, William Stein <wst...@gmail.com> wrote:
>
> > Q: What's the difference between a duck?
> > A: A duck's got feathers.
>
> "Elephants are interesting and useful. Feathers are interesting
> and useful.
> Elephants with feathers are a curiosity. Is SAGE an elephant
> with feathers?
>
> And the future is pretty grim."
>
>
>
I just re-read that post. I think the question is relevant. I have
heard from several people who installed just Maxima, after seeing that
Sage had just too much irrelevant stuff.
I generally view "adding facilities" as something positive, but if
the merging of too much stuff means that you have to make compromises
in the design, maybe not.
A guess would be that coercions would cause problems.
Or that too much detail would cause problems ... e.g. an engineer
might not want to specify algebraic domains -- something that made
Axiom unattractive, I think. Maybe this affects Sage?
And my assessment today (grim?) is about the same. Certainly I'm
stubborn, and that's a factor. Doubting the effectiveness of python
and open-source as a way to build a quality core system is a factor.
Ondrej Certik
One observation I made in this thread is that it is vastly everyone
against rjf. I will not judge who is right or wrong, but I think if
rjf was not responding, this sage-flame list would be mostly dead! So
I thank again everyone for discussing.
Ondrej
rjf
On Sep 22, 11:21 am, William Stein <wst...@gmail.com> wrote:
> It's not exactly /dev/null. Any member can read the full archives.
>
> William
counts.
Presumably the discussion would not be indexed by Google etc.
no free beer, no free speech here, anyway.
Bye
RJF
Tom Boothby
> I guess you don't really believe in open source when it really
> counts.
> Presumably the discussion would not be indexed by Google etc.
> no free beer, no free speech here, anyway.
Nobody promised you free beer, did they? Quite a number of Sage
developers are tea-totalers -- well, I'm a coffee-totaler, personally.
I guess that means you can have my beer, and it won't cost you a
cent. But no, there has never been any promise of free speech on
these forums. You'll note that sage-devel is moderated, and has spam
filters. Some consider commercial speech to be free speech, but you
haven't been too noisy about that, have you?
Anyway, "open source" means different things to different people.
Some prefer "virulently open" like RMS, others prefer "free for any
use" like myself. The former tend to equate open source with free
speech; or even some sort of fundamental right to benefit from the
work of others. I equate open source with sharing because once I've
written the code to do what I want it to do, I'm happy to have my code
and you can have it too if you want.
And now this board has gone into latter, more "passive" form. People
can get on and read all the dirt they want. But, so everybody can
feel comfortable being themselves, we've set it up so one won't
stumble upon the muck we sling in here, just searching for our names
on google. Indeed, William attempted to encourage neutrality of the
forum by adding you and I both as moderators. I certainly don't
intend to block anybody but spammers, and I trust you to do the same.
So what's the problem?
> Bye
See you all too soon, I'm sure.
--tom
William Stein
-- William
> I will not judge who is right or wrong, but I think if
> rjf was not responding, this sage-flame list would be mostly dead! So
> I thank again everyone for discussing.
>
> Ondrej
>
> >
>
William Stein
>
> On Tue, Sep 22, 2009 at 10:55 PM, rjf <fat...@gmail.com> wrote:
>
>> I guess you don't really believe in open source when it really
>> counts.
>> Presumably the discussion would not be indexed by Google etc.
>> no free beer, no free speech here, anyway.
>
> Nobody promised you free beer, did they? Quite a number of Sage
around, I would be happy to buy you a free beer!
-- William
> developers are tea-totalers -- well, I'm a coffee-totaler, personally.
> I guess that means you can have my beer, and it won't cost you a
> cent. But no, there has never been any promise of free speech on
> these forums. You'll note that sage-devel is moderated, and has spam
> filters. Some consider commercial speech to be free speech, but you
> haven't been too noisy about that, have you?
>
> Anyway, "open source" means different things to different people.
> Some prefer "virulently open" like RMS, others prefer "free for any
> use" like myself. The former tend to equate open source with free
> speech; or even some sort of fundamental right to benefit from the
> work of others. I equate open source with sharing because once I've
> written the code to do what I want it to do, I'm happy to have my code
> and you can have it too if you want.
>
> And now this board has gone into latter, more "passive" form. People
> can get on and read all the dirt they want. But, so everybody can
> feel comfortable being themselves, we've set it up so one won't
> stumble upon the muck we sling in here, just searching for our names
> on google. Indeed, William attempted to encourage neutrality of the
> forum by adding you and I both as moderators. I certainly don't
> intend to block anybody but spammers, and I trust you to do the same.
> So what's the problem?
>
>> Bye
>
> See you all too soon, I'm sure.
>
> --tom
>
> >
>
Ondrej Certik
>
> On Tue, Sep 22, 2009 at 11:23 PM, Tom Boothby <tomas....@gmail.com> wrote:
>>
>> On Tue, Sep 22, 2009 at 10:55 PM, rjf <fat...@gmail.com> wrote:
>>
>>> I guess you don't really believe in open source when it really
>>> counts.
>>> Presumably the discussion would not be indexed by Google etc.
>>> no free beer, no free speech here, anyway.
>>
>> Nobody promised you free beer, did they? Quite a number of Sage
>
> Actually, I'll be in Berkeley the first week in November. If you're
> around, I would be happy to buy you a free beer!
Very cool!
Ondrej
Harald Schilly
> That's your current feeling. How good are you at predicting the
> future?
on that feeling i'm pretty sure that i'll stick to that :)
h
Marshall Hampton
Q: What's the difference between a duck?
-Marshall