experimental packages

2 views
Skip to first unread message

Uri

unread,
Aug 2, 2010, 9:58:23 AM8/2/10
to sage-devel
Hi,
I'd like to add an experimental package, but I can't find the
procedure to do it anywhere. Do I have to open a Trac ticket? Or just
post a message here with a link to the .spkg? Is a description of the
package needed?
Thanks in advance!

Uri

Dr. David Kirkby

unread,
Aug 2, 2010, 10:43:25 AM8/2/10
to sage-...@googlegroups.com

I would personally consider doing everything required of a standard package,
with the exception of changing the install and deps files to make it standard.

http://www.sagemath.org/doc/developer/producing_spkgs.html#creating-a-new-spkg

I don't think opening a trac ticket will do any harm. Just make it clear it's
experimental.

Dave

David Joyner

unread,
Aug 2, 2010, 10:45:24 AM8/2/10
to sage-...@googlegroups.com
On Mon, Aug 2, 2010 at 9:58 AM, Uri <oriol.c...@gmail.com> wrote:
> Hi,
> I'd like to add an experimental package, but I can't find the
> procedure to do it anywhere.

Some info is here:
http://www.sagemath.org/doc/developer/disseminating_code.html


> Do I have to open a Trac ticket?

Yes please.

> Or just post a message here with a link to the .spkg?


Please do that too.


> Is a description of the package needed?


Yes please, as well as the machine(s) that the spkg
installs on.


> Thanks in advance!
>
> Uri
>
> --
> To post to this group, send an email to sage-...@googlegroups.com
> To unsubscribe from this group, send an email to sage-devel+...@googlegroups.com
> For more options, visit this group at http://groups.google.com/group/sage-devel
> URL: http://www.sagemath.org
>

Dr. David Kirkby

unread,
Aug 2, 2010, 10:58:03 AM8/2/10
to sage-...@googlegroups.com
On 08/ 2/10 03:43 PM, Dr. David Kirkby wrote:

> I don't think opening a trac ticket will do any harm. Just make it clear
> it's experimental.

I should add to that, it would be good if the difference between experimental
and optional could be clarified. The implication from reading the web pages is
that experimental is more unstable, but nowhere can I find what are the
requirements for each. Some people just put their packages as optional, without
going through any experimental stage.

Dave

William Stein

unread,
Aug 2, 2010, 12:00:11 PM8/2/10
to sage-...@googlegroups.com

Could you formulate some more precise descriptions of experimental and
optional, for discussion?

>
> Dave


>
> --
> To post to this group, send an email to sage-...@googlegroups.com
> To unsubscribe from this group, send an email to
> sage-devel+...@googlegroups.com
> For more options, visit this group at
> http://groups.google.com/group/sage-devel
> URL: http://www.sagemath.org
>

--
William Stein
Professor of Mathematics
University of Washington
http://wstein.org

John H Palmieri

unread,
Aug 2, 2010, 1:33:37 PM8/2/10
to sage-devel
On Aug 2, 9:00 am, William Stein <wst...@gmail.com> wrote:
> On Mon, Aug 2, 2010 at 7:58 AM, Dr. David Kirkby
>
> <david.kir...@onetel.net> wrote:
> > On 08/ 2/10 03:43 PM, Dr. David Kirkby wrote:
>
> >> I don't think opening a trac ticket will do any harm. Just make it clear
> >> it's experimental.
>
> > I should add to that, it would be good if the difference between
> > experimental and optional could be clarified. The implication from reading
> > the web pages is that experimental is more unstable, but nowhere can I find
> > what are the requirements for each. Some people just put their packages as
> > optional, without going through any experimental stage.
>
> Could you formulate some more precise descriptions of experimental and
> optional, for discussion?

I think that, among other things, optional packages should build on
all "supported platforms" (whatever that means -- this phrase needs to
be defined carefully somewhere: see #9487), while experimental ones
need not.

--
John

kcrisman

unread,
Aug 2, 2010, 1:52:16 PM8/2/10
to sage-devel

> > Could you formulate some more precise descriptions of experimental and
> > optional, for discussion?
>
> I think that, among other things, optional packages should build on
> all "supported platforms" (whatever that means -- this phrase needs to
> be defined carefully somewhere: see #9487), while experimental ones
> need not.

+1 - but do our current optional packages do so?

(Or that they build out of the box on most supported platforms, with
specific instructions for compiling for newer ones?)

- kcrisman

William Stein

unread,
Aug 2, 2010, 1:54:53 PM8/2/10
to sage-...@googlegroups.com
On Mon, Aug 2, 2010 at 10:52 AM, kcrisman <kcri...@gmail.com> wrote:
>
>> > Could you formulate some more precise descriptions of experimental and
>> > optional, for discussion?
>>
>> I think that, among other things, optional packages should build on
>> all "supported platforms" (whatever that means -- this phrase needs to
>> be defined carefully somewhere: see #9487), while experimental ones
>> need not.
>
> +1 - but do our current optional packages do so?

Definitely not. Ensuring this would be a massive project, which would
be well worth doing.

I used to at least test regularly that they installed on sage.math,
but I haven't even done that lately.
Regularly ensuring at least that all the optional spkg's install
*somewhere* would be a step forward from the
current situation.


>
> (Or that they build out of the box on most supported platforms, with
> specific instructions for compiling for newer ones?)
>
> - kcrisman
>

kcrisman

unread,
Aug 2, 2010, 2:17:26 PM8/2/10
to sage-devel


On Aug 2, 1:54 pm, William Stein <wst...@gmail.com> wrote:
> On Mon, Aug 2, 2010 at 10:52 AM, kcrisman <kcris...@gmail.com> wrote:
>
> >> > Could you formulate some more precise descriptions of experimental and
> >> > optional, for discussion?
>
> >> I think that, among other things, optional packages should build on
> >> all "supported platforms" (whatever that means -- this phrase needs to
> >> be defined carefully somewhere: see #9487), while experimental ones
> >> need not.
>
> > +1 - but do our current optional packages do so?
>
> Definitely not.  Ensuring this would be a massive project, which would
> be well worth doing.
>
> I used to at least test regularly that they installed on sage.math,
> but I haven't even done that lately.
> Regularly ensuring at least that all the optional spkg's install
> *somewhere* would be a step forward from the
> current situation.
>

We need to ask the R people how they do this. Somehow, they have a
system that ensures that packages are not accepted as optional
packages on CRAN unless they build on some build bot - I don't know
any of the details, but it sounds like the difference between the CRAN
and http://www.rforge.net/, which specifically does not have "make
check on-commit, nighlty builds of packages, testing on various
plarforms and full CRAN-like reposity access". The R foundation is
also a lot bigger (and it's an official FSF project), so maybe that is
how they do it.

- kcrisman

kcrisman

unread,
Aug 2, 2010, 2:19:40 PM8/2/10
to sage-devel


On Aug 2, 2:17 pm, kcrisman <kcris...@gmail.com> wrote:
> On Aug 2, 1:54 pm, William Stein <wst...@gmail.com> wrote:
>
>
>
>
>
> > On Mon, Aug 2, 2010 at 10:52 AM, kcrisman <kcris...@gmail.com> wrote:
>
> > >> > Could you formulate some more precise descriptions of experimental and
> > >> > optional, for discussion?
>
> > >> I think that, among other things, optional packages should build on
> > >> all "supported platforms" (whatever that means -- this phrase needs to
> > >> be defined carefully somewhere: see #9487), while experimental ones
> > >> need not.
>
> > > +1 - but do our current optional packages do so?
>
> > Definitely not.  Ensuring this would be a massive project, which would
> > be well worth doing.
>
> > I used to at least test regularly that they installed on sage.math,
> > but I haven't even done that lately.
> > Regularly ensuring at least that all the optional spkg's install
> > *somewhere* would be a step forward from the
> > current situation.
>
> We need to ask the R people how they do this.  Somehow, they have a
> system that ensures that packages are not accepted as optional
> packages on CRAN unless they build on some build bot - I don't know
> any of the details, but it sounds like the difference between the CRAN
> andhttp://www.rforge.net/, which specifically does not have "make

Sorry, I seem to have misread that - this also seems to have these as
features, but perhaps not requirements for use, as (apparently)
opposed to CRAN.

Harald Schilly

unread,
Aug 2, 2010, 2:41:33 PM8/2/10
to sage-devel
On Aug 2, 4:58 pm, "Dr. David Kirkby" <david.kir...@onetel.net> wrote:
> .. if the difference between experimental
> and optional could be clarified.

Personally. if I could define it, I would use these definitions:

1. standard: that's well tested, included in each distribution,
essential, and the responsibility of the Sage project that they work.

2. optional: that's some addition to sage that is supported by the
sage project (more or less), but if it builds on all systems or not is
not an issue. It should follow some defined standards, everybody who
wants to improve one of it has to go through trac+review to do so and
there should be a matrix on which systems one of these packages is
expected to work. If requested, some of them may be promoted to
"standard", as it already happened more than once.

3. experimental (which I would rename to "contributed"): this is a set
of packages a user has created, it is not supported by the sage
project, but only by one or more maintainers who are the only ones who
are responsible for it. they might not work on all systems, they might
not work at all, but most importantly, they bypass the trac+review
process.

The reason why I wish there is this 3rd category is that for example
the openopt package that i've created, it is up for review (#7708) for
more than 6 months. That's longer than the release cycle. If it won't
work somewhere, I would try to fix it and take the responsibility if
it is wrong or faulty. I don't see the benefit of still providing a
really outdated version of it (officially) while not pushing this
updated version.

H

William Stein

unread,
Aug 5, 2010, 12:24:59 AM8/5/10
to sage-...@googlegroups.com

I think I'm personally happy with the above definitions.

William

David Kirkby

unread,
Aug 5, 2010, 2:28:28 AM8/5/10
to sage-...@googlegroups.com
On 5 August 2010 05:24, William Stein <wst...@gmail.com> wrote:
> On Mon, Aug 2, 2010 at 11:41 AM, Harald Schilly
> <harald....@gmail.com> wrote:
>> On Aug 2, 4:58 pm, "Dr. David Kirkby" <david.kir...@onetel.net> wrote:
>>> .. if the difference between experimental
>>> and optional could be clarified.
>>
>> Personally. if I could define it, I would use these definitions:
>>
>> 1. standard: that's well tested, included in each distribution,
>> essential, and the responsibility of the Sage project that they work.

Agreed.

>> 2. optional: that's some addition to sage that is supported by the
>> sage project (more or less), but if it builds on all systems or not is
>> not an issue.

Like most, if not all others who responded, I feel optional shold
build on all systems, though I would make an exception for packages
like valgrind, which are inherriently non-portable and would require a
monumental effort to make them portable.

To my knowledge, all of Mathematica's optional packages work on all
releases of Mathematica.

>> 3. experimental (which I would rename to "contributed"): this is a set
>> of packages a user has created, it is not supported by the sage
>> project, but only by one or more maintainers who are the only ones who
>> are responsible for it. they might not work on all systems, they might
>> not work at all, but most importantly, they bypass the trac+review
>> process.

Personally I think there should be a record on trac of even
experimental packages. Not a review process - just a record of the
date the package was requested to be made experimental and the date
that it actually got put on the Sage web site.

In any case, someone is going to have to load the package onto the web
site, and add a short description to a web page. The contents of that
description should be put on the trac ticket.

> I think I'm personally happy with the above definitions.
>
> William

Me two, except for the two points above..

Dave

Reply all
Reply to author
Forward
Message has been deleted
0 new messages