tcl-core — Tcl Core Team discussions about Tcl development

Re: [TCLCORE] New TIP #706: Expose three Tk "In Context" functions via stubs table

[Marc C. <cul...@gm...>]

Note that subpixel values are an essential feature when a Retina display is
used on a Mac.  All screen coordinates are floats in macOS, and the unit is
what Apple calls a "point" - not to be confused with what Adobe calls a
"point" (1 / 72 inches)  and also not to be confused with what printers
have called a "point" for hundreds of years (in America,  1 / 72.27 inches).

A screen pixel on a Retina display has size 0.5 Apple points.

- Marc

On Sun, Dec 1, 2024 at 8:15 AM Benjamin Riefenstahl <
b.r...@tu...> wrote:

> Hi Kevin, all.
>
> "Kevin Walzer" writes:
> > I'm curious - why are these functions needed on macOS and not
> > X11/Windows?
>
> Interesting to see that these are still around ;-)
>
> At the time that we invented these functions, the Mac OS X text renderer
> used sub-pixel calculations for displaying characters.  The actual
> measuring APIs gave fixed-point numbers, not integers.  This meant that
> measuring multiple ranges of a string using the Tk functions and adding
> the results gave serious rounding errors, which showed up in the display
> of the text widget.  Using the InContext-functions gives pixel values
> that add up to the same sums as the text renderer and avoids those
> problems.
>
> Note that the text rendering was subsequently rewritten to use a
> different API, and I do not know how the new renderer compares with the
> old one.  I do not have a Mac any more so I can not check.
>
> Incidentally, InContext-rendering and -measuring also avoids problems
> with text shaping, which is used e.g. with Indic scripts, where glyphs
> of completely different width may be used for the same character,
> depending on the context.  This would also be useful on X11/Windows.
>
> HTH, benny
>
>
> _______________________________________________
> Tcl-Core mailing list
> Tcl...@li...
> https://lists.sourceforge.net/lists/listinfo/tcl-core
>

Re: [TCLCORE] New TIP #706: Expose three Tk "In Context" functions via stubs table

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

Op za 30 nov 2024 om 16:53 schreef Francois Vogel:

> Hi all,
>
> This is to let you know that I have written a new TIP:
>
>      https://core.tcl-lang.org/tips/doc/trunk/tip/706.md
>
> The goal is to expose 3 Tk functions in the stubs table. This is needed
> to let rtext (the revised text widget as a TEA package) build and link
> under macOS aqua. About rtext:
>

Let's do this for Tk 9.0.1 (and 8.7). This should have been part
of TIP #580:
    Rationale:
       Those exports are meant to be used by the revised_text widget...

sorry that I missed it.

Regards,
     Jan Nijtmans

Re: [TCLCORE] New TIP #706: Expose three Tk "In Context" functions via stubs table

[Paul O. <pa...@po...>]

I would also like to see it in 9.0.1.
Tested Tk and rtext branch tip-706 on several systems with no problems.

Paul


Am 01.12.2024 um 03:24 schrieb apnmbx-public--- via Tcl-Core:
> With the caveat that I am not very knowledgeable about Tk internals, I'd be inclined to include this in 9.0.1.
>
> It is not new functionality as such, just making existing functions public, and afaict does not expose internal implementation details.
>
> I expect 9.0.1 to be the first Tcl/Tk 9 release in wide deployment and the "base" against which most Tcl/Tk 9 extensions will be written so it makes sense to me to maximize the stubs interface in the release without waiting on 9.1
>
> /Ashok
>
> -----Original Message-----
> From: Francois Vogel <fvo...@fr...>
> Sent: Saturday, November 30, 2024 9:33 PM
> To: tcl...@li...
> Subject: Re: [TCLCORE] New TIP #706: Expose three Tk "In Context" functions via stubs table
>
> The one thing I'm wondering is whether this may be merged in the current
> trunk branch (for Tk 9.0.1) or if it should wait for 9.1. Please take
> field "Tk-version" of the TIP as a prompt for discussion.
>
> F.
>
> Le 30/11/2024 à 16:52, Francois Vogel a écrit :
>> Hi all,
>>
>> This is to let you know that I have written a new TIP:
>>
>>      https://core.tcl-lang.org/tips/doc/trunk/tip/706.md
>>
>> The goal is to expose 3 Tk functions in the stubs table. This is
>> needed to let rtext (the revised text widget as a TEA package) build
>> and link under macOS aqua. About rtext:
>>
>>      https://chiselapp.com/user/fvogel/repository/rtext
>>
>> Comments, thoughts, all welcome. Depending on the feedback I get, vote
>> will be triggered in possibly less than a week.
>>
>> Regards,
>>
>> Francois
>>
>>
>>
>> _______________________________________________
>> Tcl-Core mailing list
>> Tcl...@li...
>> https://lists.sourceforge.net/lists/listinfo/tcl-core
>
> _______________________________________________
> Tcl-Core mailing list
> Tcl...@li...
> https://lists.sourceforge.net/lists/listinfo/tcl-core
>
>
>
> _______________________________________________
> Tcl-Core mailing list
> Tcl...@li...
> https://lists.sourceforge.net/lists/listinfo/tcl-core

Re: [TCLCORE] Proposed addition of a new Tcl_Obj internal rep field for pointer+size

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

Op zo 1 dec 2024 om 15:50 schreef apnmbx-public--- via Tcl-Core:

> I would therefore like to add the following additional field to the
> Tcl_ObjIntenalRep union
>
>
>
> struct {
>
>    void *ptr;
>
>    Tcl_Size size;
>
> } ptrAndSize;
>

If we do that, the "localVarName", "substcode" and "bignum" objtypes,
should be changed to use the new type. Currently they mis-use the
twoPtrValue.ptr2 value to store an integer. That means also:  there are
already 3 use-cases in the core which support your request.

I'm OK doing this for 9.0.1 (and 8.7). Doing this with a TIP has the
advantage that we can declare ptrAndLongRep as deprecated, in
favor of the new type. I wouldn't remove ptrAndLongRep yet:
TWAPI still uses it (that would be the 4th use-case for the new type)

Hope this helps,
       Jan Nijtmans

[TCLCORE] Proposed addition of a new Tcl_Obj internal rep field for pointer+size

[<apn...@ya...>]

Custom Tcl_Obj's often require a pointer and length internal representation.
In Tcl 8, the Tcl_Obj.internalRep.ptrAndLongRep field targeted this use
case.

 

With Tcl 9, where supported lengths on 64-bit platforms are greater than
ULONG_MAX on Windows, this no longer suffices for the purpose. Hijacking the
twoPtrValue field involves casts, not desirable imo.

 

I would therefore like to add the following additional field to the
Tcl_ObjIntenalRep union

 

struct {

   void *ptr;

   Tcl_Size size;

} ptrAndSize;

 

Note Tcl_Size was chosen, not Tcl_WideInt,  ptrdiff_t or size_t so that the
size of Tcl_Obj is not affected on any platform for both 32- and 64-bit Tcl
9 and 8.7.

 

I would like to see this for 9.0.1.

 

Anybody have objections?

 

Is this something that would need a TIP and a vote?

 

/Ashok 

Re: [TCLCORE] New TIP #706: Expose three Tk "In Context" functions via stubs table

[Benjamin R. <b.r...@tu...>]

Hi Kevin, all.

"Kevin Walzer" writes:
> I'm curious - why are these functions needed on macOS and not
> X11/Windows?

Interesting to see that these are still around ;-)

At the time that we invented these functions, the Mac OS X text renderer
used sub-pixel calculations for displaying characters.  The actual
measuring APIs gave fixed-point numbers, not integers.  This meant that
measuring multiple ranges of a string using the Tk functions and adding
the results gave serious rounding errors, which showed up in the display
of the text widget.  Using the InContext-functions gives pixel values
that add up to the same sums as the text renderer and avoids those
problems.

Note that the text rendering was subsequently rewritten to use a
different API, and I do not know how the new renderer compares with the
old one.  I do not have a Mac any more so I can not check.

Incidentally, InContext-rendering and -measuring also avoids problems
with text shaping, which is used e.g. with Indic scripts, where glyphs
of completely different width may be used for the same character,
depending on the context.  This would also be useful on X11/Windows.

HTH, benny

Re: [TCLCORE] New TIP #706: Expose three Tk "In Context" functions via stubs table

[Francois V. <fvo...@fr...>]

In Tk, the macOS platform is the only one for which TK_DRAW_IN_CONTEXT 
is defined, leading to the InContext measuring/drawing functions being 
called. I can't tell why exactly, it must be due to mac-specificities of 
some kind, it's a different way of measuring/drawing characters than on 
other platforms.

On Windows and Linux, Tk does not use any InContext functions(*). No 
context is actually used when measuring or drawing a string. If they 
would be called, the InContext functions just call the out-of-context 
variants (I have documented this in my proposed implementation in the 
tip-706 branch).

/(*) Tk_UnderlineCharsInContext() actually is called by 
Tk_UnderlineChars() on all platforms, however 
Tk_UnderlineCharsInContext() just makes two calls to 
Tk_MeasureCharsInContext(), which on Linux and Windows is 
Tk_MeasureChars(), so the statement "On Windows and Linux, Tk does not 
use any InContext functions" can really be seen as true./

F.

Le 01/12/2024 à 03:54, Marc Culler a écrit :
> I don't know the answer, but I suspect that they are needed on all 
> platforms but only (currently) used on macOS.  The difference from the 
> ones without the context in the name seems to be how end of line 
> markers are handled.
>
> - Marc
>
> On Sat, Nov 30, 2024 at 8:35 PM Kevin Walzer <kw...@co...> wrote:
>
>     I'm curious - why are these functions needed on macOS and not
>     X11/Windows?
>
>     On 11/30/24 10:52 AM, Francois Vogel wrote:
>     > Hi all,
>     >
>     > This is to let you know that I have written a new TIP:
>     >
>     > https://core.tcl-lang.org/tips/doc/trunk/tip/706.md
>     >
>     > The goal is to expose 3 Tk functions in the stubs table. This is
>     > needed to let rtext (the revised text widget as a TEA package)
>     build
>     > and link under macOS aqua. About rtext:
>     >
>     > https://chiselapp.com/user/fvogel/repository/rtext
>     >
>     > Comments, thoughts, all welcome. Depending on the feedback I
>     get, vote
>     > will be triggered in possibly less than a week.
>     >
>     > Regards,
>     >
>     > Francois
>     >
>     >
>     >
>     > _______________________________________________
>     > Tcl-Core mailing list
>     > Tcl...@li...
>     > https://lists.sourceforge.net/lists/listinfo/tcl-core
>     _______________________________________________
>     Tcl-Core mailing list
>     Tcl...@li...
>     https://lists.sourceforge.net/lists/listinfo/tcl-core
>
>
>
> _______________________________________________
> Tcl-Core mailing list
> Tcl...@li...
> https://lists.sourceforge.net/lists/listinfo/tcl-core

Re: [TCLCORE] New TIP #706: Expose three Tk "In Context" functions via stubs table

[Marc C. <cul...@gm...>]

I don't know the answer, but I suspect that they are needed on all
platforms but only (currently) used on macOS.  The difference from the ones
without the context in the name seems to be how end of line markers are
handled.

- Marc

On Sat, Nov 30, 2024 at 8:35 PM Kevin Walzer <kw...@co...> wrote:

> I'm curious - why are these functions needed on macOS and not X11/Windows?
>
> On 11/30/24 10:52 AM, Francois Vogel wrote:
> > Hi all,
> >
> > This is to let you know that I have written a new TIP:
> >
> >     https://core.tcl-lang.org/tips/doc/trunk/tip/706.md
> >
> > The goal is to expose 3 Tk functions in the stubs table. This is
> > needed to let rtext (the revised text widget as a TEA package) build
> > and link under macOS aqua. About rtext:
> >
> >     https://chiselapp.com/user/fvogel/repository/rtext
> >
> > Comments, thoughts, all welcome. Depending on the feedback I get, vote
> > will be triggered in possibly less than a week.
> >
> > Regards,
> >
> > Francois
> >
> >
> >
> > _______________________________________________
> > Tcl-Core mailing list
> > Tcl...@li...
> > https://lists.sourceforge.net/lists/listinfo/tcl-core
> _______________________________________________
> Tcl-Core mailing list
> Tcl...@li...
> https://lists.sourceforge.net/lists/listinfo/tcl-core
>

Re: [TCLCORE] New TIP #706: Expose three Tk "In Context" functions via stubs table

[Kevin W. <kw...@co...>]

<div><img width="1" height="1" src='https://fedbdhd.r.tsp1-brevo.net/tr/op/ksR2bt79eKm5g5xZxgE8c6PJ1hdx87KAPiq10XagL36AV1SL4PzPmbLZw2ZedilD6XiA_v5iODJMFw4o_ze0WvstlsIk7toPWuKQyjncFt-ct7hW19JGNXEdVL36JzAiT8q6-ebyG86Pls_2tk4B59qh_bbB5w0N_5vcBtUo5rWXDkd4jDScN4jgRIz_lG3-0BkEm5egRgF8LJqIhPXdBep5z8C-' /></div>I'm curious - why are these functions needed on macOS and not X11/Windows?<br/><br/>On 11/30/24 10:52 AM, Francois Vogel wrote:<br/>> Hi all,<br/>><br/>> This is to let you know that I have written a new TIP:<br/>><br/>>     https://core.tcl-lang.org/tips/doc/trunk/tip/706.md<br/>><br/>> The goal is to expose 3 Tk functions in the stubs table. This is <br/>> needed to let rtext (the revised text widget as a TEA package) build <br/>> and link under macOS aqua. About rtext:<br/>><br/>>     https://chiselapp.com/user/fvogel/repository/rtext<br/>><br/>> Comments, thoughts, all welcome. Depending on the feedback I get, vote <br/>> will be triggered in possibly less than a week.<br/>><br/>> Regards,<br/>><br/>> Francois<br/>><br/>><br/>><br/>> _______________________________________________<br/>> Tcl-Core mailing list<br/>> Tcl...@li...<br/>> https://lists.sourceforge.net/lists/listinfo/tcl-core<br/>

Re: [TCLCORE] New TIP #706: Expose three Tk "In Context" functions via stubs table

[<apn...@ya...>]

With the caveat that I am not very knowledgeable about Tk internals, I'd be inclined to include this in 9.0.1.

It is not new functionality as such, just making existing functions public, and afaict does not expose internal implementation details.

I expect 9.0.1 to be the first Tcl/Tk 9 release in wide deployment and the "base" against which most Tcl/Tk 9 extensions will be written so it makes sense to me to maximize the stubs interface in the release without waiting on 9.1

/Ashok

-----Original Message-----
From: Francois Vogel <fvo...@fr...> 
Sent: Saturday, November 30, 2024 9:33 PM
To: tcl...@li...
Subject: Re: [TCLCORE] New TIP #706: Expose three Tk "In Context" functions via stubs table

The one thing I'm wondering is whether this may be merged in the current 
trunk branch (for Tk 9.0.1) or if it should wait for 9.1. Please take 
field "Tk-version" of the TIP as a prompt for discussion.

F.

Le 30/11/2024 à 16:52, Francois Vogel a écrit :
> Hi all,
>
> This is to let you know that I have written a new TIP:
>
>     https://core.tcl-lang.org/tips/doc/trunk/tip/706.md
>
> The goal is to expose 3 Tk functions in the stubs table. This is 
> needed to let rtext (the revised text widget as a TEA package) build 
> and link under macOS aqua. About rtext:
>
>     https://chiselapp.com/user/fvogel/repository/rtext
>
> Comments, thoughts, all welcome. Depending on the feedback I get, vote 
> will be triggered in possibly less than a week.
>
> Regards,
>
> Francois
>
>
>
> _______________________________________________
> Tcl-Core mailing list
> Tcl...@li...
> https://lists.sourceforge.net/lists/listinfo/tcl-core


_______________________________________________
Tcl-Core mailing list
Tcl...@li...
https://lists.sourceforge.net/lists/listinfo/tcl-core

Re: [TCLCORE] State of TclPro Debugger

[<apn...@ya...>]

A port of the debugger to Tcl 9 (needed extensive changes for namespace variable resolution) is at https://github.com/apnadkarni/TclProDebug (originally forked from Flightaware).

 

Not a user myself and have done very minimal testing with *Tcl*. It has an issue with Tk scripts (see https://sourceforge.net/p/magicsplat/tickets/20/), I think having to do with msgcat changes, that I have not had chance to look at yet.

 

Volunteers to become “official” maintainers would be welcomed!

 

/Ashok

 

 

 

From: Steve Landers <st...@di...> 
Sent: Sunday, December 1, 2024 4:18 AM
To: Tcl List Core <tcl...@li...>; Gustaf Neumann (sslmail) <ne...@wu...>
Subject: Re: [TCLCORE] State of TclPro Debugger

 

In addition to the GiHub repo Andreas mentions at https://github.com/activestate/tdk there are a couple of additional fork worth looking at:  the Flightaware one at https://github.com/activestate/tdk and a more recent fork of that at https://github.com/puremourning/TclProDebug

 

-- Steve

On 1 Dec 2024 at 1:02 AM +0800, Gustaf Neumann (sslmail) <ne...@wu... <mailto:ne...@wu...> >, wrote:



Dear Core Team,

 

I am working towards the next NaviServer 5 release, which is a change in the major version. This deserves also a major cleanup. While working through the documentation, I stepped over NaviServer’s support for the TclPro debugger. NaviServer has a command “ns_adp_debug” which allows connecting to this debugger. 

 

Since I have no access to the debugger, I have no means to verify, whether the interface still works or the NaviServer is correct.

Does it still exist? How can i get access to this? If this is dead software, it should be dropped from the forthcoming release (it is still contained in the older releases).

 

Any hints?

The Tcl pages [1] state "The easiest and fastest way to debug Tcl applications!"

 

all the best

-g

 

 

[1] https://www.tcl.tk/software/tclpro/debugger.html

_______________________________________________
Tcl-Core mailing list
Tcl...@li... <mailto:Tcl...@li...> 
https://lists.sourceforge.net/lists/listinfo/tcl-core

Re: [TCLCORE] State of TclPro Debugger

[Christian W. <Chr...@t-...>]

On 11/30/2024 10:18 PM, Andreas Kupries wrote:

> ...
> Shortly ActiveState stopped managing the TDK, which was shortly after
> I got moved to HP, all the sources became open, see:
>
> 	https://github.com/activestate/tdk
> ...

and LUCKily, the olden 8.6 stuff which I mentioned in my last posting
contains most of TDK ready for immediate consumption, e.g.

   ./vanillawish-linux64 builtin:TDK/debugger

Cheers,
C.

Re: [TCLCORE] State of TclPro Debugger

[Steve L. <st...@di...>]

In addition to the GiHub repo Andreas mentions at https://github.com/activestate/tdk there are a couple of additional fork worth looking at:  the Flightaware one at https://github.com/activestate/tdk and a more recent fork of that at https://github.com/puremourning/TclProDebug

-- Steve
On 1 Dec 2024 at 1:02 AM +0800, Gustaf Neumann (sslmail) <ne...@wu...>, wrote:
> Dear Core Team,
>
> I am working towards the next NaviServer 5 release, which is a change in the major version. This deserves also a major cleanup. While working through the documentation, I stepped over NaviServer’s support for the TclPro debugger. NaviServer has a command “ns_adp_debug” which allows connecting to this debugger.
>
> Since I have no access to the debugger, I have no means to verify, whether the interface still works or the NaviServer is correct.
> Does it still exist? How can i get access to this? If this is dead software, it should be dropped from the forthcoming release (it is still contained in the older releases).
>
> Any hints?
> The Tcl pages [1] state "The easiest and fastest way to debug Tcl applications!"
>
> all the best
> -g
>
>
> [1] https://www.tcl.tk/software/tclpro/debugger.html
> _______________________________________________
> Tcl-Core mailing list
> Tcl...@li...
> https://lists.sourceforge.net/lists/listinfo/tcl-core

Re: [TCLCORE] State of TclPro Debugger

[Andreas K. <and...@gm...>]

> Dear Core Team,
 
> I am working towards the next NaviServer 5 release, which is a
> change in the major version. This deserves also a major
> cleanup. While working through the documentation, I stepped over
> NaviServer’s support for the TclPro debugger. NaviServer has a
> command “ns_adp_debug” which allows connecting to this debugger.

After Scriptics went down the TclPro suite of tools became the
ActiveState TclDevKit suite of tools, containing the debugger as well.

Shortly ActiveState stopped managing the TDK, which was shortly after
I got moved to HP, all the sources became open, see:

	https://github.com/activestate/tdk

> Since I have no access to the debugger, I have no means to verify,
> whether the interface still works or the NaviServer is correct.
> Does it still exist? How can i get access to this? If this is dead
> software, it should be dropped from the forthcoming release (it is
> still contained in the older releases).
 
> Any hints?
> The Tcl pages [1] state "The easiest and fastest way to debug Tcl applications!"
 
> all the best
> -g
  
> [1] https://www.tcl.tk/software/tclpro/debugger.html

-- 
Happy Tcling,
	Andreas Kupries <and...@gm...>
			<https://core.tcl-lang.org/akupries/>
			<https://akupries.tclers.tk/>
	Developer @	SUSE Software Solutions Germany GmbH
-------------------------------------------------------------------------------

[TCLCORE] bug-a8e4f76ce4 (MemoryModule, Win32)

[Christian W. <Chr...@t-...>]

Howdy,

(for background see ticket https://core.tcl-lang.org/tcl/info/a8e4f76ce498ad64)

right now I have updated the LUCK stuff (see https://wiki.tcl-lang.org/page/LUCK)
including an early state of Jan's work regarding the above mentioned ticket.

Since undroidwish, vanillatclsh, and vanillawish, which are the basis of LUCK,
contain many DLLs, now is time to get an early impression on how (un)stable,
(un)suitable, (un)testable, (un)realistic this approach of loading native code
(without littering the native filesystem) might be, finally.

This is no current shiny new 9.x tech but based on good ole 8.6. Anyways should
it be sufficient to demonstrate the idea.

Cheers,
chw

Re: [TCLCORE] New TIP #706: Expose three Tk "In Context" functions via stubs table

[nicolas b. <sl1...@gm...>]

Hi,
the branch builds fine on macOS15

++

Le sam. 30 nov. 2024 à 17:03, Francois Vogel <fvo...@fr...> a écrit :

> The one thing I'm wondering is whether this may be merged in the current
> trunk branch (for Tk 9.0.1) or if it should wait for 9.1. Please take
> field "Tk-version" of the TIP as a prompt for discussion.
>
> F.
>
> Le 30/11/2024 à 16:52, Francois Vogel a écrit :
> > Hi all,
> >
> > This is to let you know that I have written a new TIP:
> >
> >     https://core.tcl-lang.org/tips/doc/trunk/tip/706.md
> >
> > The goal is to expose 3 Tk functions in the stubs table. This is
> > needed to let rtext (the revised text widget as a TEA package) build
> > and link under macOS aqua. About rtext:
> >
> >     https://chiselapp.com/user/fvogel/repository/rtext
> >
> > Comments, thoughts, all welcome. Depending on the feedback I get, vote
> > will be triggered in possibly less than a week.
> >
> > Regards,
> >
> > Francois
> >
> >
> >
> > _______________________________________________
> > Tcl-Core mailing list
> > Tcl...@li...
> > https://lists.sourceforge.net/lists/listinfo/tcl-core
>
>
> _______________________________________________
> Tcl-Core mailing list
> Tcl...@li...
> https://lists.sourceforge.net/lists/listinfo/tcl-core
>

[TCLCORE] State of TclPro Debugger

[Gustaf N. (sslmail) <ne...@wu...>]

Dear Core Team,

I am working towards the next NaviServer 5 release, which is a change in the major version. This deserves also a major cleanup. While working through the documentation, I stepped over NaviServer’s support for the TclPro debugger. NaviServer has a command “ns_adp_debug” which allows connecting to this debugger. 

Since I have no access to the debugger, I have no means to verify, whether the interface still works or the NaviServer is correct.
Does it still exist? How can i get access to this? If this is dead software, it should be dropped from the forthcoming release (it is still contained in the older releases).

Any hints?
The Tcl pages [1] state "The easiest and fastest way to debug Tcl applications!"

all the best
-g


[1] https://www.tcl.tk/software/tclpro/debugger.html

Re: [TCLCORE] New TIP #706: Expose three Tk "In Context" functions via stubs table

[Francois V. <fvo...@fr...>]

The one thing I'm wondering is whether this may be merged in the current 
trunk branch (for Tk 9.0.1) or if it should wait for 9.1. Please take 
field "Tk-version" of the TIP as a prompt for discussion.

F.

Le 30/11/2024 à 16:52, Francois Vogel a écrit :
> Hi all,
>
> This is to let you know that I have written a new TIP:
>
>     https://core.tcl-lang.org/tips/doc/trunk/tip/706.md
>
> The goal is to expose 3 Tk functions in the stubs table. This is 
> needed to let rtext (the revised text widget as a TEA package) build 
> and link under macOS aqua. About rtext:
>
>     https://chiselapp.com/user/fvogel/repository/rtext
>
> Comments, thoughts, all welcome. Depending on the feedback I get, vote 
> will be triggered in possibly less than a week.
>
> Regards,
>
> Francois
>
>
>
> _______________________________________________
> Tcl-Core mailing list
> Tcl...@li...
> https://lists.sourceforge.net/lists/listinfo/tcl-core

[TCLCORE] New TIP #706: Expose three Tk "In Context" functions via stubs table

[Francois V. <fvo...@fr...>]

Hi all,

This is to let you know that I have written a new TIP:

     https://core.tcl-lang.org/tips/doc/trunk/tip/706.md

The goal is to expose 3 Tk functions in the stubs table. This is needed 
to let rtext (the revised text widget as a TEA package) build and link 
under macOS aqua. About rtext:

     https://chiselapp.com/user/fvogel/repository/rtext

Comments, thoughts, all welcome. Depending on the feedback I get, vote 
will be triggered in possibly less than a week.

Regards,

Francois

Re: [TCLCORE] TIP 705 - Affirm Tcl License

[Pietro C. <ga...@ga...>]

On Nov 29 2024, 15:59 +0000, Kevin Kenny <kev...@gm...> wrote:
[-- Type: text/plain; charset=UTF-8, Encoding: quoted-printable, Size: 2.8K --]
>On Wed, Nov 27, 2024 at 3:41 AM Pietro Cerutti via Tcl-Core <
>tcl...@li...> wrote:
>
>> 3. As a non-US citizen, I'm having a real hard time understanding the
>> "GOVERNMENT USE" clause. I probably understand that it doesn't apply to
>> me, but if I am to license any of my code with a license that has such a
>> clause, I guess I'm better off understanding what it's all about.
>> Is this still relevant / necessary / ever desired for new code?
>>
>
>It has never been relevant/necessary.  It goes back to the days when Tcl
>was developed at Sun.
>
>At the time, open-source software was rather an unfamiliar concept. There
>were a great many government contracts that required all work product to be
>turned over to the government, including assignment of patents and
>copyrights.  There was an exception in the rules for building atop
>commercial software - nobody really thought that the procurement rules
>would require you to hand over someone else's operating system, compiler or
>runtime library! But, as I said, open source software was not well
>understood, and there were even speculations (fueled by some software
>companies' whispers) that it might be unlawful to use it in government
>work.
>
>This was not helped because DFARS (Defense Federal Acquisition RegulationS)
>at the time used the term 'sold commercial software', rather than a neutral
>term like 'third-party software'.  There was no concept that there might be
>software owned by a third party that anyone had the right to use without
>paying! (So technically there was an argument that it could not be used in
>government work: it wasn't sold, but could not be conveyed exclusively to
>the government!)
>
>There was some low-level procurement officer that was uncomfortable, and
>Sun preemptively added this clause to try to clarify the situation. When
>the legalese is parsed, it says that Tcl is software owned by another party
>that cannot be conveyed to government ownership, but that the government
>has the same right to use and modify it as anyone else. Which (as the
>legalities have sorted out) would have been the case with or without the
>'US GOVERNMENT USERS' clause.

Thanks for the interesting historical insight!

>But we don't have the opportunity to change the license on the existing
>code.  Someone still owns those copyrights - and given the dissolution of
>several of the companies on the list and the requirement to prove chain of
>copyright ownership by written records, it will be inordinately difficult
>to determine who that 'someone' is and get them all to agree to a change of
>the terms.  Delete that clause, and there's too much chance that some
>spurious claimant to emerge and wreak legal havoc, possibly on a completely
>innocent company with Tcl embedded in its product.  The claim doesn't need
>to be meritorious to be devastating.

Far from me to propose to relicense existing code! In my opening 
paragraph I have explicitely written "new code" for this reason :)

-- 
Pietro Cerutti
I have pledged to give 10% of income to effective charities
and invite you to join me - https://givingwhatwecan.org

Re: [TCLCORE] TIP 705 - Affirm Tcl License

[Kevin K. <kev...@gm...>]

On Wed, Nov 27, 2024 at 3:41 AM Pietro Cerutti via Tcl-Core <
tcl...@li...> wrote:

> 3. As a non-US citizen, I'm having a real hard time understanding the
> "GOVERNMENT USE" clause. I probably understand that it doesn't apply to
> me, but if I am to license any of my code with a license that has such a
> clause, I guess I'm better off understanding what it's all about.
> Is this still relevant / necessary / ever desired for new code?
>

It has never been relevant/necessary.  It goes back to the days when Tcl
was developed at Sun.

At the time, open-source software was rather an unfamiliar concept. There
were a great many government contracts that required all work product to be
turned over to the government, including assignment of patents and
copyrights.  There was an exception in the rules for building atop
commercial software - nobody really thought that the procurement rules
would require you to hand over someone else's operating system, compiler or
runtime library! But, as I said, open source software was not well
understood, and there were even speculations (fueled by some software
companies' whispers) that it might be unlawful to use it in government
work.

This was not helped because DFARS (Defense Federal Acquisition RegulationS)
at the time used the term 'sold commercial software', rather than a neutral
term like 'third-party software'.  There was no concept that there might be
software owned by a third party that anyone had the right to use without
paying! (So technically there was an argument that it could not be used in
government work: it wasn't sold, but could not be conveyed exclusively to
the government!)

There was some low-level procurement officer that was uncomfortable, and
Sun preemptively added this clause to try to clarify the situation. When
the legalese is parsed, it says that Tcl is software owned by another party
that cannot be conveyed to government ownership, but that the government
has the same right to use and modify it as anyone else. Which (as the
legalities have sorted out) would have been the case with or without the
'US GOVERNMENT USERS' clause.

But we don't have the opportunity to change the license on the existing
code.  Someone still owns those copyrights - and given the dissolution of
several of the companies on the list and the requirement to prove chain of
copyright ownership by written records, it will be inordinately difficult
to determine who that 'someone' is and get them all to agree to a change of
the terms.  Delete that clause, and there's too much chance that some
spurious claimant to emerge and wreak legal havoc, possibly on a completely
innocent company with Tcl embedded in its product.  The claim doesn't need
to be meritorious to be devastating.

It's ironic, but the law sometimes makes it harder to give something away
than to sell it.

-- 
73 de ke9tv/2, Kevin

Re: [TCLCORE] Confusion around TCL_WIDE_INT_IS_LONG in Tcl 8 vs Tcl 9

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

Op do 28 nov 2024 om 11:40 schreef Harald Oehlmann:

> Hi Ashok,
> that looks interesting. What I read from the TIP, the correct the check
> would be:
>
> #if defined(TCL_WIDE_INT_IS_LONG) && TCL_WIDE_INT_IS_LONG = 1
>

No, that's not correct. In Tcl 9.0 (and 8.7 as well)
TCL_WIDE_INT_IS_LONG is defined on 64-bit platforms,
undefined 32-bit platforms. It still can still be used to test
if LONG is 64-bit or 32-bit. That's how it is used inside Tcl

Hope this helps,
      Jan Nijtmans

Re: [TCLCORE] Should recent clam fix go to Tk 8.6.15 release

[Csaba N. <csa...@t-...>]

IMHO, it should.  It just fixes a small optical inconsistency related to 
the disabled state.  The fact that it was committed into the 
core-8-6-branch had exactly the goal to see it incorporated into the 
future 8.6 releases.  8.6.16 will be the first future 8.6 release.

Best regards,

Csaba


Am 28.11.24 um 12:15 schrieb Harald Oehlmann:
> Dear Tk team,
> could a Wizard please give advice to Don, if the bugfix for
> https://core.tcl-lang.org/tk/info/a69fd7cdc7
> should go to 8.6.16 release ?
> 
> Thanks,
> Harald
> 
> 
> _______________________________________________
> Tcl-Core mailing list
> Tcl...@li...
> https://lists.sourceforge.net/lists/listinfo/tcl-core

-- 
Csaba Nemethi  https://www.nemethi.de  mailto:csa...@t-...

Re: [TCLCORE] Should recent clam fix go to Tk 8.6.15 release

[François V. <fvo...@fr...>]

I'd say yes.Regards,François 
-------- Message d'origine --------De : Harald Oehlmann <har...@el...> Date : 28/11/2024  12:16  (GMT+01:00) À : Tcl Core List <tcl...@li...> Objet : [TCLCORE] Should recent clam fix go to Tk 8.6.15 release Dear Tk team,could a Wizard please give advice to Don, if the bugfix forhttps://core.tcl-lang.org/tk/info/a69fd7cdc7should go to 8.6.16 release ?Thanks,Harald

[TCLCORE] Should recent clam fix go to Tk 8.6.15 release

[Harald O. <har...@el...>]

Dear Tk team,
could a Wizard please give advice to Don, if the bugfix for
https://core.tcl-lang.org/tk/info/a69fd7cdc7
should go to 8.6.16 release ?

Thanks,
Harald