Re: [TCLCORE] TIP 716 ready for comments

[da S. P. J <pet...@fl...>]

I will be so happy to see the backside of Tcl_DStringToExternal and its hyena cousins. ☺

From: Jan Nijtmans <jan...@gm...>
Date: Friday, April 25, 2025 at 02:38
To: tcl...@li... <tcl...@li...>
Subject: Re: [TCLCORE] TIP 716 ready for comments
Op di 15 apr 2025 om 21: 02 schreef apnmbx-public--- via Tcl-Core <tcl-core@ lists. sourceforge. net>: > The rules are (as per TIP 716): > > Use the Unicode Windows API as Tcl/Tk already do. Then these issues do not arise. > If


Op di 15 apr 2025 om 21:02 schreef apnmbx-public--- via Tcl-Core

<tcl...@li...>:

> The rules are (as per TIP 716):

>

> Use the Unicode Windows API as Tcl/Tk already do. Then these issues do not arise.

> If you have to use ANSI API’s, use Windows encoding settings, not Tcl encoding settings.



So, let's establish that TIP #716 changes the rules!



I did some investigation which extensions will be broken by TIP #716.

But - first -

let's remember the rules how they were in  Tcl 8.x.



1) Use the Unicode Windows API as Tcl/Tk already do. Then these issues

do not arise:

       Tcl_WinUtfToTChar(str, strlen(str), &ds);

       _wfopen((WCHAR *)Tcl_DStringValue(ds), L"r")



2) If you have to use ANSI API’s, use the following:

       Tcl_DStringInit(&ds);

       Tcl_DStringToExtenal(NULL, str, -1, &ds);

       fopen(Tcl_DStringValue(ds), "r")



3) If you only care about ASCII characters, just do:

       fopen(str, "r")



In Tcl 9.0.0/9.0.3, all 3 rules start working correctly with UTF-8

(except when using

the GDI API but there is no known extension using that). Whatever

choice is made, 1), 2) or 3), all work fine with Tcl 9.0.0/9.0.1.  If

TIP #716 is

accepted, 2) and 3) won't work correctly with characters outside ASCII any more.

Which extensions will suffer from that? I found the following:



    blt/src/bltWinImage.c (line 1194)

    expect/exp_main_sub.c (line 813)

    pikchr/pikchr.c (line 8136)

    tkimg/sgi/sgi.c (line 1512)

    VecTcl/WavReader/generic/wavreader.c (line 88)



So, as long as BLT images, pikchr, sgi or waveform filenames

use ASCII characters only, nothing to worry about. When running

in Tcl 9.0.0 or 9.0.1, nothing to worry about either. But with

TIP #716, the rules change, and 2) or 3) can no longer handle

characters outside ASCII.



Hope this helps,

        Jan Nijtmans





_______________________________________________

Tcl-Core mailing list

Tcl...@li...

https://urldefense.com/v3/__https://lists.sourceforge.net/lists/listinfo/tcl-core__;!!MvWE!FTxd301TfyO5GgcGfdBSuSnO-jh1562z8eQNudPLkWdmxnKFL9fojX7lFMF-seoyoO-rIpaQa8FobgbaN2F45Mu0k43tNA$<https://urldefense.com/v3/__https:/lists.sourceforge.net/lists/listinfo/tcl-core__;!!MvWE!FTxd301TfyO5GgcGfdBSuSnO-jh1562z8eQNudPLkWdmxnKFL9fojX7lFMF-seoyoO-rIpaQa8FobgbaN2F45Mu0k43tNA$>