Tcl8.5 roadmap
si...@whiteowl.co.uk
and many minor.
Is there a plan for releasing 8.5? With some major new functionality
(e.g. dict, tile, clock re-write) it seems to me that it would be good
to get it into mainstream use as it is this year and plan for an 8.6
next year to contain those major enhancements (OO framework?) that
aren't yet ready.
One of the great things about tcl is the community effort that goes
into developing the language, but as well as enabling people to enhance
the language the changes really do need to be released to customers.
8.5 seems to be growing without any overall guidance and should
probably be cut at its current state and moved into beta with release
pencilled in for October/November this year. Features not yet ready
could then be put into 8.6 for release at the end of 2007.
What do other people think?
Simon Geard
Michael Schlenker
I basically agree that there should be a release this year, even if not
all unfinished TIPs can go in (even if it is sad for some i would like
to see like, OO support, IPv6 Support, Named Colors for Tk, zlib
support, tkpng merge into Tk, and some others).
But the best thing to really get a release is to participate in the work
to get a release.
1. Implement one of the open TIPs
2. Fix one of the open Bugs and supply a patch, or provide helpful
comments or testing for open bugs
3. Test the latest alpha release and report bugs (especially test the
release candidates if they come up, to have bugs fixed before one of the
rare releases !)
4. Write new tutorial lesson for the 8.5 features in the tcltutorial
4a. Create a nice stylesheet for the tutorial
5. Help with maintaining/enhancing the tcl.tk webpage
6. Write cool demos for the new 8.5 features to be included with Tcl or Tk
Some of those would remove burden from the maintainers, some would just
bring stalled TIPs forward, some would be really nice to have and good
for introducing new users to the power of Tcl/Tk.
Michael
Donal K. Fellows
> There are many great things already implemented for v8.5 - some major
> and many minor. Is there a plan for releasing 8.5?
Not a meaningful one; such plans as there have been in the past have
also been known to be more wishlists for release schedules...
> With some major new functionality
> (e.g. dict, tile, clock re-write) it seems to me that it would be good
> to get it into mainstream use as it is this year and plan for an 8.6
> next year to contain those major enhancements (OO framework?) that
> aren't yet ready.
[...]
> What do other people think?
I think I'll be very interested to see what other people think. :-)
Donal.
Robert Hicks
si...@whiteowl.co.uk wrote:
<snip>
> What do other people think?
>
> Simon Geard
Well I know Tile is going in there so we will have to wait for that. I
agree that we should get 8.5 out the door with or without the OO stuff.
I would rather whoever is doing that not get rushed into implementing
something they weren't comfortable with.
Is there a list of open TIPS for 8.5 somewhere?
:Robert
Eckhard Lehmann
Robert Hicks wrote:
> si...@whiteowl.co.uk wrote:
> <snip>
> > What do other people think?
> >
> > Simon Geard
>
> Well I know Tile is going in there so we will have to wait for that. I
> agree that we should get 8.5 out the door with or without the OO stuff.
> I would rather whoever is doing that not get rushed into implementing
> something they weren't comfortable with.
Basically I have the same opinion.
But as for the OO: When 8.5 is released without OO, this will probably
be available not until 8.6 to the public. And when the release of 8.6
takes as long as the release of 8.5, the public will have to wait
another 2-3 years or longer for a *builtin* object system. This is bad,
IMHO, as it kicks the language behind.
I really think, that the OO stuff should be finished for 8.5 if
possible, and the release of 8.5 should be delayed for as long as this
takes. We're waiting a long time for 8.5 now - so 3-6 months more or
less don't really matter.
Eckhard
Donal K. Fellows
> Well I know Tile is going in there so we will have to wait for that. I
> agree that we should get 8.5 out the door with or without the OO stuff.
> I would rather whoever is doing that not get rushed into implementing
> something they weren't comfortable with.
Err, that's me. And the problem is purely that my work workload has
been stupid the past year or so. :-(
> Is there a list of open TIPS for 8.5 somewhere?
Several places. http://tip.tcl.tk/ is an obvious starting point, and
http://wiki.tcl.tk/12753 is another good spot to look.
Donal.
Robert Hicks
I was not slandering you! lol
I was just making a point. 8.5.X could be 8.5 + OO + fixes for all I
care. I just don't want to rush it and it have flaws.
:Robert
Robert Hicks
add OO at 8.5.1 or 8.5.X.
However, this is all speculation since I am not in the know for these
things.
:Robert
Bryan Oakley
> Robert Hicks wrote:
>
>>Well I know Tile is going in there so we will have to wait for that. I
>>agree that we should get 8.5 out the door with or without the OO stuff.
>>I would rather whoever is doing that not get rushed into implementing
>>something they weren't comfortable with.
>
>
> Err, that's me. And the problem is purely that my work workload has
> been stupid the past year or so. :-(
>
Is there anything any of us can do specifically to help get the OO TIP
implemented? I have some time to write some man pages and maybe some
coding or testing or something.
--
Bryan Oakley
http://www.tclscripting.com
Donal K. Fellows
Bryan Oakley wrote:
> Is there anything any of us can do specifically to help get the OO TIP
> implemented? I have some time to write some man pages and maybe some
> coding or testing or something.
There's only one really tough part to do, and that's the method
dispatch engine. I'm still totally stalled on that, with various
attempts that at best don't work properly. Get that right and the rest
of it is not much more than data structure management.
Help is welcome. Try the tip-257-implementation-branch branch in the
Tcl CVS tree. If you want write permission but don't have it, terms are
the usual Tcl Maintainer ones, i.e. we will own your firstborn's soul!
(Or something like that. ;-) )
Donal.
Larry W. Virden
Robert Hicks wrote:
> You don't have to wait for an 8.6. We could get 8.5 out the door and
> add OO at 8.5.1 or 8.5.X.
changes to the underlying major APIs... the stubs tables. And I would
expect the OO stuff to change those. So if the OO stuff doesn't go into
8.5, it would seem, to me, that they would probably need to go into
8.6. HOWEVER, I don't think we need to wait another 2-3 years for 8.6.
7.5 -> 7.6 was 6 months or so.
7.6->8.0 was 10 months
8.0->8.1 was 20 months or so
8.1->8.2 was about 4 months
8.2->8.3 was about 6 months
8.3->8.4 was 31 months
8.4->8.5 has been 45 months...
Robert Hicks
Hey that is cool. I didn't start using Tcl until 8.4 so I didn't have
any historical arguments. So there you go. We wouldn't have to wait for
8.6. : )
:Robert
Eckhard Lehmann
> Larry W. Virden wrote:
> > The transition between 8.4 to 8.5, and then 8.5 to 8.6, has to do with
> > changes to the underlying major APIs... the stubs tables. And I would
> > expect the OO stuff to change those. So if the OO stuff doesn't go into
> > 8.5, it would seem, to me, that they would probably need to go into
> > 8.6.
Anyway, I don't think it is a good idea to introduce new features in
patchlevel releases. By definition they are made for bug fixes...
Robert Hicks wrote:
> Hey that is cool. I didn't start using Tcl until 8.4 so I didn't have
> any historical arguments. So there you go. We wouldn't have to wait for
> 8.6. : )
Imo, there is another point to consider. Although I favour incremental
updates, I think that the release of 8.5 should be well prepared - for
marketing reasons. Such a release can be a big chance to make a product
more popular if it is announced as "even faster than before" and
"packed with a load of new features" etc (OO, [dict] ...). People like
to hear these kind of buzz. If the release is coupled to a schedule and
some marketing/non-tech work (website, blog-planet, magazine articles,
even banner advertisements and some kind of merchandise, etc.), they
would pay attention and see that Tcl moves towards the future with big
steps.
As I can see, 8.5 really makes a big step in terms of functionality
compared to previous versions. So I think that this kind of marketing
with the release would pay off...
Eckhard
Robert Hicks
Eckhard Lehmann wrote:
<snip>
>
> Imo, there is another point to consider.
Point well taken.
:Robert
si...@whiteowl.co.uk
original post. Even if it were decided that 8.5 would be released on
the current feature set it would still take most of the rest of the
year to turn it into a full release. I am not critisizing anyone for
this - I know that people have day job, families and other commitments
that necessarily limit their time. However I do think that there should
now be some real focus on releasing 8.5 as it is rather than continuing
to let its feature set grow. I suggest the following for inclusion:
1) all completed TIP's
2) tile - is a new feature and can therefore be added as it is close to
the release date. As someone else pointed out it is new and therefore
shouldn't upset any existing code.
3) OO - would be wonderful if it were ready but I don't think 8.5
should be held up for it - 8.6 could follow in 6 months or so when it's
ready.
In order to give developers time to finalize their code I'd suggest 1st
September 2006 as the drop-dead date by which any new features will be
accepted. The whole bug-fixing, release preparation cycle can be given
real focus.
Simon Geard
Donal K. Fellows
This is true pretty much by definition; a TIP is completed (Final) once
it is implemented, tested, documented and in the CVS trunk.
> 2) tile - is a new feature and can therefore be added as it is close to
> the release date. As someone else pointed out it is new and therefore
> shouldn't upset any existing code.
> 3) OO - would be wonderful if it were ready but I don't think 8.5
> should be held up for it - 8.6 could follow in 6 months or so when it's
> ready.
These two are the big remaining features we want to add before 8.5.
Pressure Joe English and Jeff Hobbs over the first, and me over the second.
Donal.
Larry W. Virden
Larry W. Virden wrote:
> 7.5 -> 7.6 was 6 months or so.
> 7.6->8.0 was 10 months
> 8.0->8.1 was 20 months or so
> 8.1->8.2 was about 4 months
> 8.2->8.3 was about 6 months
> 8.3->8.4 was 31 months
> 8.4->8.5 has been 45 months...
There is a context however that may not be obvious. With the exception
of the 8.0 to 8.1 transition, things for quite a number of years was
progressing with major releases 2-3 times a year. I know that, in my
own case, managers grew tired of scheduling time for developers to have
to deal with changes necessary to make the transitions and to keep up.
So I seem to recall that, a number of years ago, there was discussion
that the minor releases would continue regularly, getting bug fixes,
etc. out, but that the major releases, which might cause lots of change
for a developer, were going to occur less frequently, resulting,
hopefully, in a perception of the stability that the tcl code already
displayed.
ped...@west.com
Add tiles and OO.
Release that as 9.0
Work on 8.6 and 9.1 bug fixes and updates.
There needs to be more "tcl news". Places like lwn.net have been
posting dr.dobbs links for the last 40 months with no other big tcl
news.
3 or 4 versions in the next 6 months sounds like an "active language"
to me.
Michael Schlenker
> Release 8.5 right now.
> Add tiles and OO.
> Release that as 9.0
> Work on 8.6 and 9.1 bug fixes and updates.
>
> There needs to be more "tcl news". Places like lwn.net have been
> posting dr.dobbs links for the last 40 months with no other big tcl
> news.
stuff, graphics or writing good texts and some free time at hand.
Some ideas, add your own or just do something...
- Where is the RoR style video showing the world how easy it is to write
network servers in Tcl?
- Start a tile theming contest for Tk, offer some prices
- Run a blog and discuss Tcl/Tk and add references wherever appropriate
- Write an article about event based programming and feature Tcl
- Write cool earth shaking apps with Tcl...
> 3 or 4 versions in the next 6 months sounds like an "active language"
> to me.
Sounds like an immature language to me. One release every six month
would be good though, 50-60 months for a release is a too much.
Michael
Bryan Oakley
> Not hard to generate news, just needs someone good with public relations
> stuff, graphics or writing good texts and some free time at hand.
>
> Some ideas, add your own or just do something...
> - Run a blog and discuss Tcl/Tk and add references wherever appropriate
.text insert end "shameless plug for http://www.tclscripting.com"
Not exactly a daily blog, and not (yet?) a weekly blog, but I'm working
on it.
Michael A. Cleverly
> .text insert end "shameless plug for http://www.tclscripting.com"
>
> Not exactly a daily blog, and not (yet?) a weekly blog, but I'm working
> on it.
Bryan, I've really enjoyed reading your articles. Very educational & well
written. Keep up the good work!
Michael
ped...@west.com
Also I thought the point of making Tk an independent package, was to
allow the Tcl Core to be released more often...
Ian Gay
$H71....@newssvr13.news.prodigy.com:
> ht
A nice site! But why the reduced font size? Truly, I have my browser
default font set the way I like it. Why do so many sites nowadays
force you to do a font resize first thing?
Ian
--
*********** To reply by e-mail, make w single in address
**************
Bryan Oakley
> Bryan Oakley <oak...@bardo.clearlight.com> wrote in news:sNYjg.97331
> $H71....@newssvr13.news.prodigy.com:
>
>
>>ht
>
>
> A nice site! But why the reduced font size? Truly, I have my browser
> default font set the way I like it. Why do so many sites nowadays
> force you to do a font resize first thing?
>
> Ian
>
I didn't realize it had a reduced font size. It wasn't intentional on my
part. I hate those web sites that choose tiny fonts too. I'll admit,
though, I'm blindly using a template that came with my web site
software. I'll dig in to the template and see what I can figure out.
What OS and web browser are you using?
Ian Gay
Seamonkey 1.02 on winXP SP2. It only goes about 1 step smaller, not
as nasty as some sites I could mention...
WL
Ian Gay <g...@sfuu.ca> wrote:
>Bryan Oakley <oak...@bardo.clearlight.com> wrote in
>news:449082BF...@bardo.clearlight.com:
>
>> Ian Gay wrote:
>>> Bryan Oakley <oak...@bardo.clearlight.com> wrote in
>>> news:sNYjg.97331 $H71....@newssvr13.news.prodigy.com:
>>>
>>> browser default font set the way I like it. Why do so many sites
>>> nowadays force you to do a font resize first thing?
>>
>>
>Seamonkey 1.02 on winXP SP2. It only goes about 1 step smaller, not
>as nasty as some sites I could mention...
If you go to Control Panel->Display Properties->Settings->Advanced
->General, what ist he DPI setting set to? I'm going to guess
120 DPI, or something not 96.
Part of the problem may be shaky support for various font
sizes. Search bugzilla for "font dpi" for a list... FWIW,
a lot of installers, applications, etc, have weird layouts
because of the dpi. Setting the dpi back down to 96 will
usually fix these things.
--
WL
real mail: wliao at sdf loSnPesAtarM org
(remove the uppercase letters...)
Donald Arseneau
> I didn't realize it had a reduced font size.
It's not necessarily reduced -- it could be enlarged for some users.
But it does set a specific font size:
body {
color: #555555;
font: 0.75em/1.6em "lucida grande", verdana, tahoma, sans-serif;
--
Donald Arseneau as...@triumf.ca
Eckhard Lehmann
Michael Schlenker wrote:
> - Run a blog and discuss Tcl/Tk and add references wherever appropriate
There are these feed aggregators everywhere, like
http://planet.gnome.org/, http://planet.maemo.org/ etc. - where blog
articles of various people are aggregated. They write mostly about
their involvement in various projects but also very general things.
It's often entertaining to fly over and read the one or other article.
On planet.gnome.org I like especially the "hacker gotchis", which are
the little pictures of the people who write..
That would be an idea for www.tcl.tk, isn't it?
Eckhard
Donal K. Fellows
> Also I thought the point of making Tk an independent package, was to
> allow the Tcl Core to be released more often...
I always thought it was rather political, having to do with people being
able to use Tk without Tcl (even though it seems that that's not the
best way of doing it, at least according to Jeff Hobbs's experiences
with Tk and Perl...)
Donal.
Gerhard Reithofer
I've a question, which may become a tipp.
What is the reason why the set command does not allow varargs?
Eg.
set a 1
set fl 1.0/3
set l {one two three}
could be written shorter and more "compact" as:
set a 1 fl 1.0/3 l {one two three}
I've just redefined "set" sometimes for my usage, but are there
arguments against this as "normal" feature?
Bye,
Gerhard Reithofer - Techn. EDV Reithofer - Technical software solutions
mailto:gerhard....@tech-edv.co.at http://www.tech-edv.co.at
Peierlhang 31b, A-8042 Graz, Austria, Phone +43-316/405707, Fax Ext. 15
Don Porter
> I've a question, which may become a tipp.
TIP 58 has already been rejected.
--
| Don Porter Mathematical and Computational Sciences Division |
| donald...@nist.gov Information Technology Laboratory |
| http://math.nist.gov/~DPorter/ NIST |
|______________________________________________________________________|
Gerhard Reithofer
> Gerhard Reithofer wrote:
> > I've a question, which may become a tipp.
>
> TIP 58 has already been rejected.
Thanks for this fast reply, I thought I've studied the tipps database.
Is there an explanation or discussion about rejected tipps available?
Jeff Hobbs
> Hi TCLers,
> I've a question, which may become a tipp.
>
> What is the reason why the set command does not allow varargs?
>
> Eg.
> set a 1
> set fl 1.0/3
> set l {one two three}
>
> could be written shorter and more "compact" as:
> set a 1 fl 1.0/3 l {one two three}
>
> I've just redefined "set" sometimes for my usage, but are there
> arguments against this as "normal" feature?
This is exactly what 'lassign' in 8.5 gives you.
Jeff
Alan Anderson
>
> What is the reason why the set command does not allow varargs?
If forced to choose a reason, I'd point out that [set] is defined to
return the value of the variable. *The* value. *The* variable.
> Eg.
> set a 1
> set fl 1.0/3
> set l {one two three}
>
> could be written shorter and more "compact" as:
> set a 1 fl 1.0/3 l {one two three}
You could also use the [foreach] idiom:
foreach {a fl l} {1 1.0/3 {one two three}} break
Donal K. Fellows
> Is there an explanation or discussion about rejected tipps available?
I try to record a summary of that information in the TIP history, but
the definitive answer is recorded in the archives of the tcl-core
mailing list since TCT members are always strongly encouraged to give
reasons for a NO vote.
If I remember right, #58 was rejected because it was felt that keeping
the [set] command very simple was a better principle to stick to.
Donal.
suchenwi
want.
Only, for performance reasons, better leave standard set as it is.
Here's another variation:
proc set* args {foreach {var val} $args {uplevel 1 [list set $var
$val]}}
Gerhard Reithofer
Thank's - also for all other replies.
Sometimes I use a similar solution, as I wrote in my 1st message.
rename set _set
proc set args {
foreach {var val} $args {
upvar $var lvar
_set lvar $val
}
}
But I have to provide this code in any app, otherwise I would use a
"regular" command.
@RS: I've "stolen" and "blown up" your function plotter example ;-)
http://www.tech-edv.co.at/programmierung/en/gplsw.html