Let's tell we are using GTK+

Let's tell we are using GTK+

[Terje Rosten]

Here are two patches which together let the lisp function
emacs-version tell about the presence of GTK+ support.

Please apply.

Thanks,

 - Terje



Index: src/xfns.c
===================================================================
RCS file: /cvsroot/emacs/emacs/src/xfns.c,v
retrieving revision 1.590
diff -c -r1.590 xfns.c
*** src/xfns.c	31 Jul 2003 16:10:53 -0000	1.590
--- src/xfns.c	3 Aug 2003 14:52:35 -0000
***************
*** 10886,10891 ****
--- 10886,10895 ----
  #endif /* USE_MOTIF */
  #endif /* USE_X_TOOLKIT */
  
+ #ifdef USE_GTK
+   Fprovide (intern ("gtk+-toolkit"), Qnil);
+ #endif /* USE_GTK */
+ 
    /* X window properties.  */
    defsubr (&Sx_change_window_property);
    defsubr (&Sx_delete_window_property);
Index: src/ChangeLog
===================================================================
RCS file: /cvsroot/emacs/emacs/src/ChangeLog,v
retrieving revision 1.3331
diff -c -r1.3331 ChangeLog
*** src/ChangeLog	31 Jul 2003 16:10:53 -0000	1.3331
--- src/ChangeLog	3 Aug 2003 14:52:37 -0000
***************
*** 1,3 ****
--- 1,7 ----
+ 2003-08-03  Terje Rosten <terjeros@phys.ntnu.no>
+ 
+ 	* xfns.c (syms_of_xfns): Let's tell we are using GTK+.
+ 
  2003-07-31  Jan Djärv  <jan.h.d@swipnet.se>
  
  	* xfns.c (xg_set_icon): Rewrite to compile with GTK 2.0 and 2.2.



Index: lisp/version.el
===================================================================
RCS file: /cvsroot/emacs/emacs/lisp/version.el,v
retrieving revision 2.32
diff -c -r2.32 version.el
*** lisp/version.el	4 Feb 2003 12:12:31 -0000	2.32
--- lisp/version.el	3 Aug 2003 14:53:36 -0000
***************
*** 62,67 ****
--- 62,68 ----
  		 (cond ((featurep 'motif)
  			(concat ", " (substring motif-version-string 4)))
  		       ((featurep 'x-toolkit) ", X toolkit")
+ 		       ((featurep 'gtk+-toolkit) ", GTK+ toolkit")
  		       (t ""))
  		 (if (and (boundp 'x-toolkit-scroll-bars)
  			  (memq x-toolkit-scroll-bars '(xaw xaw3d)))
Index: lisp/ChangeLog
===================================================================
RCS file: /cvsroot/emacs/emacs/lisp/ChangeLog,v
retrieving revision 1.5321
diff -c -r1.5321 ChangeLog
*** lisp/ChangeLog	2 Aug 2003 20:36:51 -0000	1.5321
--- lisp/ChangeLog	3 Aug 2003 14:53:39 -0000
***************
*** 1,3 ****
--- 1,7 ----
+ 2003-08-03 Terje Rosten <terjeros@phys.ntnu.no>
+ 
+ 	* version.el (emacs-version): Check for gtk+-toolkit.
+ 
  2003-08-02  Andreas Schwab  <schwab@suse.de>
  
  	* apropos.el (apropos-words-to-regexp): Only add `wild' if `words'

Re: Let's tell we are using GTK+

[Miles Bader]

Do people really always include the `+' in the name?  It seems sort of
superfluous and annoying in this case.  Why not just use the symbol
`gtk-toolkit' for emacs?

-Miles
-- 
Suburbia: where they tear out the trees and then name streets after them.

Re: Let's tell we are using GTK+

[Terje Rosten]

* Miles Bader
| 
| Do people really always include the `+' in the name?

I don't know.


| It seems sort of superfluous and annoying in this case.  Why not
| just use the symbol `gtk-toolkit' for emacs?

That's fine. Want a updated patch?


 - Terje

Re: Let's tell we are using GTK+

[Lute Kamstra]

Terje Rosten <terjeros@phys.ntnu.no> writes:

> Here are two patches which together let the lisp function
> emacs-version tell about the presence of GTK+ support.

[...]

> Index: lisp/version.el
> ===================================================================
> RCS file: /cvsroot/emacs/emacs/lisp/version.el,v
> retrieving revision 2.32
> diff -c -r2.32 version.el
> *** lisp/version.el	4 Feb 2003 12:12:31 -0000	2.32
> --- lisp/version.el	3 Aug 2003 14:53:36 -0000
> ***************
> *** 62,67 ****
> --- 62,68 ----
>   		 (cond ((featurep 'motif)
>   			(concat ", " (substring motif-version-string 4)))
>   		       ((featurep 'x-toolkit) ", X toolkit")
> + 		       ((featurep 'gtk+-toolkit) ", GTK+ toolkit")

Maybe the GTK version should be mentioned as well.  Handy for bug reports.

>   		       (t ""))
>   		 (if (and (boundp 'x-toolkit-scroll-bars)
>   			  (memq x-toolkit-scroll-bars '(xaw xaw3d)))

[...]


  Lute.

Re: Let's tell we are using GTK+

[Paul Jarc]

Miles Bader <miles@gnu.org> wrote:
> Do people really always include the `+' in the name?  It seems sort of
> superfluous and annoying in this case.  Why not just use the symbol
> `gtk-toolkit' for emacs?

It would be nice if the same name were used everywhere.  So in the
absence of any concrete problem, I'd prefer "gtk+", since that's the
name given to that package by its maintainers.  What makes + any more
superfluous or annoying than g, t, or k?


paul

Re: Let's tell we are using GTK+

[Terje Rosten]

* Lute Kamstra
| 
| Maybe the GTK version should be mentioned as well.

Updated patches with version information and the (annoying?) +
sign removed. emacs-version will now display something like this:

 GNU Emacs 21.3.50.1 (i686-pc-linux-gnu, GTK+ 2.2.1) of 2003-08-04 on 
 myhost.net


 - Terje


Index: src/xfns.c
===================================================================
RCS file: /cvsroot/emacs/emacs/src/xfns.c,v
retrieving revision 1.590
diff -u -r1.590 xfns.c
--- src/xfns.c	31 Jul 2003 16:10:53 -0000	1.590
+++ src/xfns.c	4 Aug 2003 19:27:05 -0000
@@ -125,6 +125,14 @@
 
 #endif /* USE_X_TOOLKIT */
 
+#ifdef USE_GTK
+
+/* GTK+ version info */
+
+static Lisp_Object Vgtk_version_string;
+
+#endif /* USE_GTK */
+
 #ifdef HAVE_X11R4
 #define MAXREQUEST(dpy) (XMaxRequestSize (dpy))
 #else
@@ -10885,6 +10893,19 @@
   Vmotif_version_string = build_string (XmVERSION_STRING);
 #endif /* USE_MOTIF */
 #endif /* USE_X_TOOLKIT */
+
+#ifdef USE_GTK
+  Fprovide (intern ("gtk-toolkit"), Qnil);
+
+  DEFVAR_LISP ("gtk-version-string", &Vgtk_version_string,
+	       doc: /* Version info for GTK+.  */);
+  {
+    char gtk_version[8];
+    sprintf (gtk_version, "%d.%d.%d", GTK_MAJOR_VERSION, GTK_MINOR_VERSION,
+	     GTK_MICRO_VERSION);
+    Vgtk_version_string = build_string (gtk_version);
+  }
+#endif /* USE_GTK */
 
   /* X window properties.  */
   defsubr (&Sx_change_window_property);
Index: src/ChangeLog
===================================================================
RCS file: /cvsroot/emacs/emacs/src/ChangeLog,v
retrieving revision 1.3331
diff -u -r1.3331 ChangeLog
--- src/ChangeLog	31 Jul 2003 16:10:53 -0000	1.3331
+++ src/ChangeLog	4 Aug 2003 19:27:08 -0000
@@ -1,3 +1,7 @@
+2003-08-04  Terje Rosten <terjeros@phys.ntnu.no>
+
+	* xfns.c (syms_of_xfns): Add information about GTK+ version.
+
 2003-07-31  Jan Djärv  <jan.h.d@swipnet.se>
 
 	* xfns.c (xg_set_icon): Rewrite to compile with GTK 2.0 and 2.2.




Index: lisp/version.el
===================================================================
RCS file: /cvsroot/emacs/emacs/lisp/version.el,v
retrieving revision 2.32
diff -u -r2.32 version.el
--- lisp/version.el	4 Feb 2003 12:12:31 -0000	2.32
+++ lisp/version.el	4 Aug 2003 19:26:46 -0000
@@ -62,6 +62,8 @@
 		 (cond ((featurep 'motif)
 			(concat ", " (substring motif-version-string 4)))
 		       ((featurep 'x-toolkit) ", X toolkit")
+		       ((featurep 'gtk-toolkit)
+			(concat ", GTK+ " gtk-version-string))
 		       (t ""))
 		 (if (and (boundp 'x-toolkit-scroll-bars)
 			  (memq x-toolkit-scroll-bars '(xaw xaw3d)))
Index: lisp/ChangeLog
===================================================================
RCS file: /cvsroot/emacs/emacs/lisp/ChangeLog,v
retrieving revision 1.5325
diff -u -r1.5325 ChangeLog
--- lisp/ChangeLog	3 Aug 2003 14:10:33 -0000	1.5325
+++ lisp/ChangeLog	4 Aug 2003 19:26:48 -0000
@@ -1,3 +1,7 @@
+2003-08-04 Terje Rosten <terjeros@phys.ntnu.no>
+
+	* version.el (emacs-version): Check for gtk-toolkit.
+
 2003-08-03  Glenn Morris  <gmorris@ast.cam.ac.uk>
 
 	* calendar/calendar.el (diary-file, diary-file-name-prefix)

Re: Let's tell we are using GTK+

[Miles Bader]

Paul Jarc wrote:
> > Do people really always include the `+' in the name?  It seems sort of
> > superfluous and annoying in this case.  Why not just use the symbol
> > `gtk-toolkit' for emacs?
>
> It would be nice if the same name were used everywhere.  So in the
> absence of any concrete problem, I'd prefer "gtk+", since that's the
> name given to that package by its maintainers.

It's not used consistently as it is; in fact, it seems to be used more
often _without_ the `+' than with it, e.g.:

  name of the library:        libgtk.so
  name of the init file:      .gtkrc
  name of include file:       <gtk-2.0/gtk/gtk.h>           (thrice!)
  name of the feature macro:  USE_GTK                       (in emacs)
  prefix used by functions:   gtk_...
  path for rc file in themes: gtk/gtkrc

Really the only place the `+' ever seems to be used is in documentation;
otherwise dropping it is apparently the convention, especially in
`technical' contexts.

> What makes + any more superfluous or annoying than g, t, or k?

Because in contexts like this (naming of symbols used in an programming
interface), punctuation tends to be be one of two things:

 (1) Noise added for readability, usually in a standard way (e.g., `_'
     in function names); usually in this case it acts basically like
     whitespace, and so tends to be ignored.

 (2) `Meaningful,' but again, used in a very stylized fashion, e.g., the
     `?' suffix used for scheme predicates, or various C conventions for
     underline prefixes.

Anything that doesn't follow these conventions ends up being harder to
use, because your brain likes to force things to fit with rules it's
used to, even when it really shouldn't.

This is especially true with `+', as it's visually not all that distinct
from other characters that _are_ used in one of the above standardized
ways (and in the case of the suggested emacs lisp symbol, the `+' was
adjacent to a `-', which makes things even more fuzzy).

-Miles
-- 
80% of success is just showing up.  --Woody Allen

Re: Let's tell we are using GTK+

[Stephen J. Turnbull]

>>>>> "Paul" == Paul Jarc <prj@po.cwru.edu> writes:

    Paul> It would be nice if the same name were used everywhere.  So
    Paul> in the absence of any concrete problem, I'd prefer "gtk+",
    Paul> since that's the name given to that package by its
    Paul> maintainers.  What makes + any more superfluous or annoying
    Paul> than g, t, or k?

Nothing.  But common practice seems to be to drop the "+".  For
example, Debian shows 251 packages containing the string "gtk", of
which only 2 contain "gtk+".  None of the library names contain the
"+".  Fink (Mac OS X) names the GTK+ library and doc packages
themselves as "gtk+", but all the packages that call GTK+ use the
shorter "gtk" (or even shorter leading "g").

So if Debian and Fink are at all representative, having the same name
everywhere favors "gtk" over "gtk+".

-- 
Institute of Policy and Planning Sciences     http://turnbull.sk.tsukuba.ac.jp
University of Tsukuba                    Tennodai 1-1-1 Tsukuba 305-8573 JAPAN
               Ask not how you can "do" free software business;
              ask what your business can "do for" free software.

Re: Let's tell we are using GTK+

[Richard Stallman]

    Do people really always include the `+' in the name?  It seems sort of
    superfluous and annoying in this case.  Why not just use the symbol
    `gtk-toolkit' for emacs?

I agree with you.

Re: Let's tell we are using GTK+

[Kenichi Handa]

In article <E19k7Gy-0002W0-O1@fencepost.gnu.org>, Richard Stallman <rms@gnu.org> writes:
>     Do people really always include the `+' in the name?  It seems sort of
>     superfluous and annoying in this case.  Why not just use the symbol
>     `gtk-toolkit' for emacs?

> I agree with you.

As "gtk" stands for "Gimp ToolKit", "gtk-toolkit" sounds
strange to me.

---
Ken'ichi HANDA
handa@m17n.org

Re: Let's tell we are using GTK+

[Masatake YAMATO]

;; Off topic

> In article <E19k7Gy-0002W0-O1@fencepost.gnu.org>, Richard Stallman <rms@gnu.org> writes:
> >     Do people really always include the `+' in the name?  It seems sort of
> >     superfluous and annoying in this case.  Why not just use the symbol
> >     `gtk-toolkit' for emacs?
> 
> > I agree with you.
> 
> As "gtk" stands for "Gimp ToolKit", "gtk-toolkit" sounds
> strange to me.

Few years ago I heard that NT of WindowsNT meant for "New Technology"
from someone.  However, I can see "Windows 2000 --- based on NT
Technology" on the screen when MS-Windows2000 is booted:-P

Masatake

Re: Let's tell we are using GTK+

[Terje Rosten]

* Kenichi Handa
| 
| As "gtk" stands for "Gimp ToolKit", "gtk-toolkit" sounds strange to
| me.

Agree. Updated patches follows.

Please apply.

Thanks,

 - Terje


src/ part:

Index: src/xfns.c
===================================================================
RCS file: /cvsroot/emacs/emacs/src/xfns.c,v
retrieving revision 1.590
diff -c -r1.590 xfns.c
*** src/xfns.c	31 Jul 2003 16:10:53 -0000	1.590
--- src/xfns.c	6 Aug 2003 07:42:47 -0000
***************
*** 125,130 ****
--- 125,138 ----
  
  #endif /* USE_X_TOOLKIT */
  
+ #ifdef USE_GTK
+ 
+ /* GTK+ version info */
+ 
+ static Lisp_Object Vgtk_version_string;
+ 
+ #endif /* USE_GTK */
+ 
  #ifdef HAVE_X11R4
  #define MAXREQUEST(dpy) (XMaxRequestSize (dpy))
  #else
***************
*** 10885,10890 ****
--- 10893,10911 ----
    Vmotif_version_string = build_string (XmVERSION_STRING);
  #endif /* USE_MOTIF */
  #endif /* USE_X_TOOLKIT */
+ 
+ #ifdef USE_GTK
+   Fprovide (intern ("gtk"), Qnil);
+ 
+   DEFVAR_LISP ("gtk-version-string", &Vgtk_version_string,
+ 	       doc: /* Version info for GTK+.  */);
+   {
+     char gtk_version[12];
+     sprintf (gtk_version, "%d.%d.%d", GTK_MAJOR_VERSION, GTK_MINOR_VERSION,
+              GTK_MICRO_VERSION);
+     Vgtk_version_string = build_string (gtk_version);
+   }
+ #endif /* USE_GTK */
  
    /* X window properties.  */
    defsubr (&Sx_change_window_property);
Index: src/ChangeLog
===================================================================
RCS file: /cvsroot/emacs/emacs/src/ChangeLog,v
retrieving revision 1.3331
diff -c -r1.3331 ChangeLog
*** src/ChangeLog	31 Jul 2003 16:10:53 -0000	1.3331
--- src/ChangeLog	6 Aug 2003 07:42:48 -0000
***************
*** 1,3 ****
--- 1,8 ----
+ 2003-08-06  Terje Rosten <terjeros@phys.ntnu.no>
+ 
+ 	* xfns.c (Vgtk_version_string): New variable.
+ 	(syms_of_xfns): DEFVAR_LISP it. Provide gtk.
+ 
  2003-07-31  Jan Djärv  <jan.h.d@swipnet.se>
  
  	* xfns.c (xg_set_icon): Rewrite to compile with GTK 2.0 and 2.2.


lisp/ part:


Index: lisp/version.el
===================================================================
RCS file: /cvsroot/emacs/emacs/lisp/version.el,v
retrieving revision 2.32
diff -c -r2.32 version.el
*** lisp/version.el	4 Feb 2003 12:12:31 -0000	2.32
--- lisp/version.el	6 Aug 2003 07:42:27 -0000
***************
*** 62,67 ****
--- 62,69 ----
  		 (cond ((featurep 'motif)
  			(concat ", " (substring motif-version-string 4)))
  		       ((featurep 'x-toolkit) ", X toolkit")
+ 		       ((featurep 'gtk)
+ 			(concat ", GTK+ " gtk-version-string))
  		       (t ""))
  		 (if (and (boundp 'x-toolkit-scroll-bars)
  			  (memq x-toolkit-scroll-bars '(xaw xaw3d)))
Index: lisp/ChangeLog
===================================================================
RCS file: /cvsroot/emacs/emacs/lisp/ChangeLog,v
retrieving revision 1.5328
diff -c -r1.5328 ChangeLog
*** lisp/ChangeLog	6 Aug 2003 01:11:11 -0000	1.5328
--- lisp/ChangeLog	6 Aug 2003 07:42:32 -0000
***************
*** 1,3 ****
--- 1,7 ----
+ 2003-08-06 Terje Rosten <terjeros@phys.ntnu.no>
+ 
+ 	* version.el (emacs-version): Check for gtk. Include gtk version info.
+ 
  2003-08-05  Richard M. Stallman  <rms@gnu.org>
  
  	* emacs-lisp/testcover.el (noreturn): Report error if does return.

Re: Let's tell we are using GTK+

[Lute Kamstra]

Terje Rosten <terjeros@phys.ntnu.no> writes:

> * Lute Kamstra
> | 
> | Maybe the GTK version should be mentioned as well.
>
> Updated patches with version information and the (annoying?) +
> sign removed. emacs-version will now display something like this:
>
>  GNU Emacs 21.3.50.1 (i686-pc-linux-gnu, GTK+ 2.2.1) of 2003-08-04 on 
>  myhost.net

I like it, but...

[...]

> +  {
> +    char gtk_version[8];
> +    sprintf (gtk_version, "%d.%d.%d", GTK_MAJOR_VERSION, GTK_MINOR_VERSION,
> +	     GTK_MICRO_VERSION);
> +    Vgtk_version_string = build_string (gtk_version);
> +  }

this seems vulnerable to a buffer overflow.  It is acceptable to use
snprintf instead, or isn't that portable enough?  (I'm not really
intimate with C.)  What could be used instead, a larger string?


  Lute.

Re: Let's tell we are using GTK+

[Terje Rosten]

* Lute Kamstra
|
| > +  {
| > +    char gtk_version[8];
| > +    sprintf (gtk_version, "%d.%d.%d", GTK_MAJOR_VERSION, GTK_MINOR_VERSION,
| > +	     GTK_MICRO_VERSION);
| > +    Vgtk_version_string = build_string (gtk_version);
| > +  }
| 
| this seems vulnerable to a buffer overflow.

That's right. In a updated patch I increased to gtk_version[12], but
that's not large enough if a GTK+ release is named e.g 2.2.cvs20030819.


| It is acceptable to use snprintf instead, or isn't that portable
| enough?  (I'm not really intimate with C.)  What could be used
| instead, a larger string?

What about using glib? If GTK+ is available then is glib available
too. A updated patch using the glib function g_snprintf is included.


 - Terje


Index: lisp/ChangeLog
===================================================================
RCS file: /cvsroot/emacs/emacs/lisp/ChangeLog,v
retrieving revision 1.5348
diff -c -r1.5348 ChangeLog
*** lisp/ChangeLog	18 Aug 2003 22:46:02 -0000	1.5348
--- lisp/ChangeLog	19 Aug 2003 10:42:16 -0000
***************
*** 1,3 ****
--- 1,8 ----
+ 2003-08-18 Terje Rosten <terjeros@phys.ntnu.no>
+ 
+ 	* version.el (emacs-version): Check for gtk. Include gtk version
+ 	info.
+ 
  2003-08-18  Luc Teirlinck  <teirllm@mail.auburn.edu>
  
  	* wid-edit.el (widget-echo-help): Make it handle expressions that
Index: lisp/version.el
===================================================================
RCS file: /cvsroot/emacs/emacs/lisp/version.el,v
retrieving revision 2.32
diff -c -r2.32 version.el
*** lisp/version.el	4 Feb 2003 12:12:31 -0000	2.32
--- lisp/version.el	19 Aug 2003 10:42:16 -0000
***************
*** 62,67 ****
--- 62,69 ----
  		 (cond ((featurep 'motif)
  			(concat ", " (substring motif-version-string 4)))
  		       ((featurep 'x-toolkit) ", X toolkit")
+ 		       ((featurep 'gtk)
+ 			(concat ", GTK+ " gtk-version-string))
  		       (t ""))
  		 (if (and (boundp 'x-toolkit-scroll-bars)
  			  (memq x-toolkit-scroll-bars '(xaw xaw3d)))
Index: src/ChangeLog
===================================================================
RCS file: /cvsroot/emacs/emacs/src/ChangeLog,v
retrieving revision 1.3335
diff -c -r1.3335 ChangeLog
*** src/ChangeLog	17 Aug 2003 00:25:37 -0000	1.3335
--- src/ChangeLog	19 Aug 2003 10:42:18 -0000
***************
*** 1,3 ****
--- 1,8 ----
+ 2003-08-18  Terje Rosten <terjeros@phys.ntnu.no>
+ 
+ 	* xfns.c (Vgtk_version_string): New variable.
+ 	(syms_of_xfns): DEFVAR_LISP it. Provide gtk.
+ 
  2003-08-16  Richard M. Stallman  <rms@gnu.org>
  
  	* editfns.c (Fencode_time): Doc fix.
Index: src/xfns.c
===================================================================
RCS file: /cvsroot/emacs/emacs/src/xfns.c,v
retrieving revision 1.590
diff -c -r1.590 xfns.c
*** src/xfns.c	31 Jul 2003 16:10:53 -0000	1.590
--- src/xfns.c	19 Aug 2003 10:42:19 -0000
***************
*** 125,130 ****
--- 125,138 ----
  
  #endif /* USE_X_TOOLKIT */
  
+ #ifdef USE_GTK
+ 
+ /* GTK+ version info */
+ 
+ static Lisp_Object Vgtk_version_string;
+ 
+ #endif /* USE_GTK */
+ 
  #ifdef HAVE_X11R4
  #define MAXREQUEST(dpy) (XMaxRequestSize (dpy))
  #else
***************
*** 10885,10890 ****
--- 10893,10911 ----
    Vmotif_version_string = build_string (XmVERSION_STRING);
  #endif /* USE_MOTIF */
  #endif /* USE_X_TOOLKIT */
+ 
+ #ifdef USE_GTK
+   Fprovide (intern ("gtk"), Qnil);
+ 
+   DEFVAR_LISP ("gtk-version-string", &Vgtk_version_string,
+                doc: /* Version info for GTK+.  */);
+   {
+     char gtk_version[12];
+     g_snprintf (gtk_version, sizeof (gtk_version), "%d.%d.%d",
+                 GTK_MAJOR_VERSION, GTK_MINOR_VERSION, GTK_MICRO_VERSION);
+     Vgtk_version_string = build_string (gtk_version);
+   }
+ #endif /* USE_GTK */
  
    /* X window properties.  */
    defsubr (&Sx_change_window_property);

Re: Let's tell we are using GTK+

[Lute Kamstra]

Terje Rosten <terjeros@phys.ntnu.no> writes:

> * Lute Kamstra
> |
> | > +  {
> | > +    char gtk_version[8];
> | > +    sprintf (gtk_version, "%d.%d.%d", GTK_MAJOR_VERSION, GTK_MINOR_VERSION,
> | > +	     GTK_MICRO_VERSION);
> | > +    Vgtk_version_string = build_string (gtk_version);
> | > +  }
> | 
> | this seems vulnerable to a buffer overflow.
>
> That's right. In a updated patch I increased to gtk_version[12], but
> that's not large enough if a GTK+ release is named e.g
> 2.2.cvs20030819.

So, let's make it 40 then, to be on the safe side.

> | It is acceptable to use snprintf instead, or isn't that portable
> | enough?  (I'm not really intimate with C.)  What could be used
> | instead, a larger string?
>
> What about using glib? If GTK+ is available then is glib available
> too. A updated patch using the glib function g_snprintf is included.

Makes sense.

Terje, did you sign copyright papers for Emacs?  If so I'll apply your
patch in a few days if nobody objects.

  Lute.

Re: Let's tell we are using GTK+

[José Roberto B. de A. Monteiro]

On Tue, Aug 19, 2003 at 11:46:21AM +0200, Lute Kamstra wrote:
> Terje Rosten <terjeros@phys.ntnu.no> writes:
> 
> > * Lute Kamstra
> > | 
> > | Maybe the GTK version should be mentioned as well.
> >
> > Updated patches with version information and the (annoying?) +
> > sign removed. emacs-version will now display something like this:
> >
> >  GNU Emacs 21.3.50.1 (i686-pc-linux-gnu, GTK+ 2.2.1) of 2003-08-04 on 
> >  myhost.net
> 
> I like it, but...
> 
> [...]
> 
> > +  {
> > +    char gtk_version[8];
> > +    sprintf (gtk_version, "%d.%d.%d", GTK_MAJOR_VERSION, GTK_MINOR_VERSION,
> > +	     GTK_MICRO_VERSION);
> > +    Vgtk_version_string = build_string (gtk_version);
> > +  }
> 
> this seems vulnerable to a buffer overflow.  It is acceptable to use
> snprintf instead, or isn't that portable enough?  (I'm not really
> intimate with C.)  What could be used instead, a larger string?

I believe, asprintf would be better, but I don't know about portability...

{
  char * gtk_version;
  asprintf (&gtk_version, "%d.%d.%d", GTK_MAJOR_VERSION, GTK_MINOR_VERSION,
          GTK_MICRO_VERSION);
  Vgtk_version_string = build_string (gtk_version);
  free(gtk_version);
}

Re: Let's tell we are using GTK+

[Betoes]

On Tue, Aug 19, 2003 at 11:46:21AM +0200, Lute Kamstra wrote:
> Terje Rosten <terjeros@phys.ntnu.no> writes:
> 
> > * Lute Kamstra
> > | 
> > | Maybe the GTK version should be mentioned as well.
> >
> > Updated patches with version information and the (annoying?) +
> > sign removed. emacs-version will now display something like this:
> >
> >  GNU Emacs 21.3.50.1 (i686-pc-linux-gnu, GTK+ 2.2.1) of 2003-08-04 on 
> >  myhost.net
> 
> I like it, but...
> 
> [...]
> 
> > +  {
> > +    char gtk_version[8];
> > +    sprintf (gtk_version, "%d.%d.%d", GTK_MAJOR_VERSION, GTK_MINOR_VERSION,
> > +	     GTK_MICRO_VERSION);
> > +    Vgtk_version_string = build_string (gtk_version);
> > +  }
> 
> this seems vulnerable to a buffer overflow.  It is acceptable to use
> snprintf instead, or isn't that portable enough?  (I'm not really
> intimate with C.)  What could be used instead, a larger string?

I believe, asprintf would be better, but I don't know about portability...

{
  char * gtk_version;
  asprintf (&gtk_version, "%d.%d.%d", GTK_MAJOR_VERSION, GTK_MINOR_VERSION,
          GTK_MICRO_VERSION);
  Vgtk_version_string = build_string (gtk_version);
  free(gtk_version);
}

Re: Let's tell we are using GTK+

[Betoes]

On Tue, Aug 19, 2003 at 11:46:21AM +0200, Lute Kamstra wrote:
> Terje Rosten <terjeros@phys.ntnu.no> writes:
> 
> > * Lute Kamstra
> > | 
> > | Maybe the GTK version should be mentioned as well.
> >
> > Updated patches with version information and the (annoying?) +
> > sign removed. emacs-version will now display something like this:
> >
> >  GNU Emacs 21.3.50.1 (i686-pc-linux-gnu, GTK+ 2.2.1) of 2003-08-04 on 
> >  myhost.net
> 
> I like it, but...
> 
> [...]
> 
> > +  {
> > +    char gtk_version[8];
> > +    sprintf (gtk_version, "%d.%d.%d", GTK_MAJOR_VERSION, GTK_MINOR_VERSION,
> > +	     GTK_MICRO_VERSION);
> > +    Vgtk_version_string = build_string (gtk_version);
> > +  }
> 
> this seems vulnerable to a buffer overflow.  It is acceptable to use
> snprintf instead, or isn't that portable enough?  (I'm not really
> intimate with C.)  What could be used instead, a larger string?

I believe, asprintf would be better, but I don't know about portability...

{
  char * gtk_version;
  asprintf (&gtk_version, "%d.%d.%d", GTK_MAJOR_VERSION, GTK_MINOR_VERSION,
          GTK_MICRO_VERSION);
  Vgtk_version_string = build_string (gtk_version);
  free(gtk_version);
}

Re: Let's tell we are using GTK+

[Andreas Schwab]

Betoes <betoes@member.fsf.org> writes:

|> I believe, asprintf would be better, but I don't know about portability...

It's a GNU extension, but a portable implementation is part of libiberty.

Andreas.

-- 
Andreas Schwab, SuSE Labs, schwab@suse.de
SuSE Linux AG, Deutschherrnstr. 15-19, D-90429 Nürnberg
Key fingerprint = 58CA 54C7 6D53 942B 1756  01D3 44D5 214B 8276 4ED5
"And now for something completely different."

Re: Let's tell we are using GTK+

[Richard Stallman]

    I believe, asprintf would be better, but I don't know about portability...

isn't asprintf a GNU libc extension?

Re: Let's tell we are using GTK+

[Simon Josefsson]

Richard Stallman <rms@gnu.org> writes:

>     I believe, asprintf would be better, but I don't know about portability...
>
> isn't asprintf a GNU libc extension?

Yes, although GNULib contain a portable stand-alone version of it,
suitable for inclusion in projects that need it.  I don't think it is
worth adding it to Emacs just to solve this problem, though, as
simpler solutions to this problem exists, at least compared to
including a printf() implementation.

Re: Let's tell we are using GTK+

[Richard Stallman]

    |> I believe, asprintf would be better, but I don't know about portability...

    It's a GNU extension, but a portable implementation is part of libiberty.

It would be ok to use this if we copy that code into Emacs,
but if it is easy to avoid, let's avoid it.

Re: Let's tell we are using GTK+

[Lute Kamstra]

Richard Stallman <rms@gnu.org> writes:

>     |> I believe, asprintf would be better, but I don't know about portability...
>
>     It's a GNU extension, but a portable implementation is part of libiberty.
>
> It would be ok to use this if we copy that code into Emacs,
> but if it is easy to avoid, let's avoid it.

It's easy to avoid it by using g_snprintf from glib, a library
required for the GTK+ build of Emacs (as suggested by Terje Rosten).
I'll commit the patch tomorrow.

  Lute.