tcl-core

[TCLCORE] Consistancy in contributed packages (Was: Tcl/Tk 8.6b1 Release Candidates)

[Jan N. <jan...@gm...>]

2008/12/19 Andreas Kupries <and...@ac...>:
>
>> Questions (that I may better understand...):
>> - Why is zlib under compat/, libtommath under /, and tdbc under pkgs?
>> Shouldn't these have very similar treatment in the location in the
>> source tree?

How about TclOO and tclZlib.c? I was wondering if they wouln't not be
served better ad contributed package. Yes, we want to make sure
that every Tcl distrubition has it, just like tdbc. TclOO is integrated
in the core, it doesn't need a package require, but provides it's own
package name and version. tclZlib.c is integrated in the core but
completely independant. It could easiliy be converted to have it's
own stub table, and be regarded as a separate extension.

Not mentioning the package names. When adding Itcl to the
contributed packages, then we will have two official bundled
packages, "tdbc" and "Itcl". Is it wise to keep the character
casing different, or should we simply provide "itcl" as well?

I think it is worth to consider this. (zlib itself and libtommath
have a clear explanation why they cannot be contributed
packages, I agree with Andreas about those two)

Regards,
          Jan Nijtmans

Re: [TCLCORE] Consistancy in contributed packages (Was: Tcl/Tk 8.6b1 Release Candidates)

[Donal K. F. <don...@ma...>]

Jan Nijtmans wrote:
> 2008/12/19 Andreas Kupries <and...@ac...>:
>>> Questions (that I may better understand...):
>>> - Why is zlib under compat/, libtommath under /, and tdbc under pkgs?
>>> Shouldn't these have very similar treatment in the location in the
>>> source tree?
> 
> How about TclOO and tclZlib.c? I was wondering if they wouln't not be
> served better ad contributed package. Yes, we want to make sure
> that every Tcl distrubition has it, just like tdbc. TclOO is integrated
> in the core, it doesn't need a package require, but provides it's own
> package name and version. tclZlib.c is integrated in the core but
> completely independant. It could easiliy be converted to have it's
> own stub table, and be regarded as a separate extension.

TclOO is deeply entwined into Tcl, and likely to become even more so.

The zlib distribution is purely there for compatibility. Our config code
really prefers the system version if that is available. OK, on Windows
we don't even check for a system version at the moment as I'm fairly
sure that Win doesn't provide one. (Am I wrong? If it's only on some
versions, then we'll probably still have to keep doing an unconditional
build of the compat code.)

I've had reports that there are problems with the msys build. I'll sort
them out later today (using techniques I've devised when working on the
Unix build rules).

Donal.

Re: [TCLCORE] Consistancy in contributed packages (Was: Tcl/Tk 8.6b1 Release Candidates)

[Donal K. F. <don...@ma...>]

Donal K. Fellows wrote:
> TclOO is deeply entwined into Tcl, and likely to become even more so.

And I forgot the main point of the message! It only exports a package
for compatibility with the standalone code that works with 8.5.

Donal.

Re: [TCLCORE] Consistancy in contributed packages (Was: Tcl/Tk 8.6b1 Release Candidates)

[Jan N. <jan...@gm...>]

2008/12/20 Donal K. Fellows <don...@ma...>:
> Donal K. Fellows wrote:
>>
>> TclOO is deeply entwined into Tcl, and likely to become even more so.
> And I forgot the main point of the message! It only exports a package
> for compatibility with the standalone code that works with 8.5.

OK, that's a clear answer. providing "TclOO" is just like an
'undocumented feature'.    :-)

However, I was talking about tclZlib.c, so the wrapper around zlib,
not zlib itself. Anyway, I'm not suggesting to change anything now
either, it's not worth delaying Tcl 8.6 for that.

Regards,
      Jan Nijtmans

Re: [TCLCORE] Consistancy in contributed packages (Was: Tcl/Tk 8.6b1 Release Candidates)

[Arnulf W. <ar...@wi...>]

Am Samstag, 20. Dezember 2008 09:33:21 schrieb Jan Nijtmans:
> 2008/12/19 Andreas Kupries <and...@ac...>:
> >> Questions (that I may better understand...):
> >> - Why is zlib under compat/, libtommath under /, and tdbc under pkgs?
> >> Shouldn't these have very similar treatment in the location in the
> >> source tree?
>
> How about TclOO and tclZlib.c? I was wondering if they wouln't not be
> served better ad contributed package. Yes, we want to make sure
> that every Tcl distrubition has it, just like tdbc. TclOO is integrated
> in the core, it doesn't need a package require, but provides it's own
> package name and version. tclZlib.c is integrated in the core but
> completely independant. It could easiliy be converted to have it's
> own stub table, and be regarded as a separate extension.
>
> Not mentioning the package names. When adding Itcl to the
> contributed packages, then we will have two official bundled
> packages, "tdbc" and "Itcl". Is it wise to keep the character
> casing different, or should we simply provide "itcl" as well?

the beta release of itcl4.0 is named "itcl" already to be consistent
with [package require itcl]

Arnulf
>
> I think it is worth to consider this. (zlib itself and libtommath
> have a clear explanation why they cannot be contributed
> packages, I agree with Andreas about those two)
>
> Regards,
>           Jan Nijtmans
>
> ---------------------------------------------------------------------------
>--- _______________________________________________
> Tcl-Core mailing list
> Tcl...@li...
> https://lists.sourceforge.net/lists/listinfo/tcl-core

[TCLCORE] Further development of LibTomMath (Was: Tcl/Tk 8.6b1 Release Candidates)

[<bo...@ma...>]

Hi there,

I'm new to the list, and may be I'm in the wrong place, if so, please tell
me.

> You wrote:
>
> **	And libtomath is an integral part of the core. We use only the
> 	sources coming with Tcl, never look for a system library, and,
> 	most important I believe, we have AFAIK forked the sources. IIRC
> 	Kevin said something to the effect that our libtommath contains
> 	a number of changes which are not in the original libtommath.
> 	IIRC the changes were given to the upstream maintainer, but
> 	rejected.

AFAIK upstream maintainer is not developing the library any more (last
minor update was December 2006). I was experimenting a bit with with Toom
multiplication and squaring using somehow more efficient sequences I
recently published ( http://ln.bodrato.it/FasterToomConvolution_pdf ).
The new code (a couple of small files) I wrote seems to be much more
efficient than the one contained in libtommath, and it should be as
portable and as readable as the original.
Are you interested in testing and maybe consider including my code in some
future releases?

Let me know,
Marco

-- 
http://bodrato.it/