tcl8.4-dev: Tcl_AppendResult

[Debian Bug Importer]

Automatically imported from Debian bug report #274212 http://bugs.debian.org/274212

[Debian Bug Importer (debzilla)]

Automatically imported from Debian bug report #274212 http://bugs.debian.org/274212

[Debian Bug Importer (debzilla)]

Message-Id: <email address hidden>
Date: Thu, 30 Sep 2004 13:41:18 +0200
From: Peter Meszaros <email address hidden>
To: Debian Bug Tracking System <email address hidden>
Subject: tcl8.4-dev: Tcl_AppendResult

Package: tcl8.4-dev
Version: 8.4.7-1
Severity: critical
Justification: breaks unrelated software

If Tcl_AppendResult appends string containing accented
charaters (iso-8859-2) then the appended string will be
trimmed at the last accented character.
Eg. (with flying chars!): in case of appending "l'al'al'al"
the result will be "l'al'al"

-- System Information:
Debian Release: 3.1
  APT prefers testing
  APT policy: (500, 'testing')
Architecture: i386 (i686)
Kernel: Linux 2.4.27
Locale: LANG=hu_HU, LC_CTYPE=hu_HU

Versions of packages tcl8.4-dev depends on:
ii libc6-dev [libc-dev] 2.3.2.ds1-16 GNU C Library: Development Librari
ii tcl8.4 8.4.7-1 Tcl (the Tool Command Language) v8

-- no debconf information

[Matt Zimmerman (mdz)]

Severity inflated, downgrading

[Chris Waters (xtifr)]

reassign 274212 tcl8.4
retitle 274212 output in latinX console drops all after last accented char
severity 274212 important
thanks

On Thu, Sep 30, 2004 at 01:41:18PM +0200, Peter Meszaros wrote:

> If Tcl_AppendResult appends string containing accented
> charaters (iso-8859-2) then the appended string will be
> trimmed at the last accented character.
> Eg. (with flying chars!): in case of appending "l'al'al'al"
> the result will be "l'al'al"

I've done some preliminary investigation. This is reproducible with
iso-8859-1 (and probably other latinX encodings). And it definitely
affects tcl8.4, not just tcl8.4-dev, so I'm going to reassign the bug
to that package. However, it seems to be related to stream output in
a text console. The string seems to be correct internally. Using the
"string length" command gives the expected value (i.e. 7 above). And
putting the string into a Tk text widget shows the entire string. And
finally, the problem can be completely avoided by using a UTF-8
environment (e.g. uxterm).

Since most uses of Tcl involve Tk, and since there's a passible
workaround even for cases that don't use Tk (use UTF-8), I'm going to
change the severity to "important". But this is still going to be one
of my top priorities. Thanks for finding and reporting this.

cheers

--
Chris Waters | Pneumonoultra- osis is too long
<email address hidden> | microscopicsilico- to fit into a single
or <email address hidden> | volcaniconi- standalone haiku

[Debian Bug Importer (debzilla)]

Message-ID: <email address hidden>
Date: Mon, 4 Oct 2004 13:40:39 -0700
From: Chris Waters <email address hidden>
To: Peter Meszaros <email address hidden>, <email address hidden>
Subject: Re: Bug#274212: tcl8.4-dev: Tcl_AppendResult

reassign 274212 tcl8.4
retitle 274212 output in latinX console drops all after last accented char
severity 274212 important
thanks

On Thu, Sep 30, 2004 at 01:41:18PM +0200, Peter Meszaros wrote:

> If Tcl_AppendResult appends string containing accented
> charaters (iso-8859-2) then the appended string will be
> trimmed at the last accented character.
> Eg. (with flying chars!): in case of appending "l'al'al'al"
> the result will be "l'al'al"

I've done some preliminary investigation. This is reproducible with
iso-8859-1 (and probably other latinX encodings). And it definitely
affects tcl8.4, not just tcl8.4-dev, so I'm going to reassign the bug
to that package. However, it seems to be related to stream output in
a text console. The string seems to be correct internally. Using the
"string length" command gives the expected value (i.e. 7 above). And
putting the string into a Tk text widget shows the entire string. And
finally, the problem can be completely avoided by using a UTF-8
environment (e.g. uxterm).

Since most uses of Tcl involve Tk, and since there's a passible
workaround even for cases that don't use Tk (use UTF-8), I'm going to
change the severity to "important". But this is still going to be one
of my top priorities. Thanks for finding and reporting this.

cheers

--
Chris Waters | Pneumonoultra- osis is too long
<email address hidden> | microscopicsilico- to fit into a single
or <email address hidden> | volcaniconi- standalone haiku

[Barry deFreese (bddebian)]

Debian maintainer is working on it.

[Francisco Isgleas (isgleas)]

My environment is set to UTF-8, and the issue persists. It shouldn't be considered an environment charset issue. If on the tcl app I do change the charset to UTF-8 I see no changes, but if set to ISO-8859-1 I get even more weird characters:

Using amsn-0.97+final-0ubuntu2
Hardy 8.04
tcl8.5.0-2
tk8.5.0-3

Expected: áéíóúü
UTF-8 setting: ´a´e´i´o´u¨u
ISO-8859-1 setting: ´a´e´i´o´u¨u

[Francisco Isgleas (isgleas)]

amsn 0.97+final-0ubuntu2
tcl8.4 8.4.16-4ubuntu1
tcl8.5 8.5.0-2
tk8.4 8.4.16-2ubuntu1
tk8.5 8.5.0-3

Weird, I have the same packages reported early, but the issue isn't present anymore. I deactivated the SCIM support on my environment, that should be the only change on configuration made on this machine. Could this be related to SCIM rather to tcl/tk?

On amsn, if I set it up to use iso8859-1, I get: áéíóí
If to UTF-8: áéíóú (expected result)
If to auto: áéíóú (expected result, as the system's environment is on UTF-8)
If to ascii: áéíóú

[Daniel T Chen (crimsun)]

Locale-based, not Tcl-based.