3:02 p.m.
This patch ...
2014-10-18 Stephen J. Turnbull <stephen(a)xemacs.org>
* fontcolor-xlike-inc.c (xft_find_charset_font):
* fontcolor-x.c (x_font_instance_truename):
* font-mgr.c (Ffc_name_unparse):
Protect Ffc_name_unparse from FcNameUnparse crash on charset property.
... caused some breakage for me. In particular, the test whether a
particular font implements a charset would always fail. The reason is
that fc patterns passed as arguments are modified in place. This patch
fixes that.
I'll push on Monday if nobody objects.
2014-10-25 Michael Sperber <mike(a)xemacs.org>
* fontcolor-x.c (x_font_instance_truename):
* font-mgr.c (Ffc_name_unparse):
* font-mgr.h (PRINT_XFT_PATTERN): Don't modify
fontconfig pattern passed in in place.
--
Regards,
Mike
diff --git a/src/font-mgr.c b/src/font-mgr.c
--- a/src/font-mgr.c
+++ b/src/font-mgr.c
＠＠ -266,9 +266,12 ＠＠
CHECK_FC_PATTERN (pattern);
/* #### Could use multiple values here to extract and return charset? */
- FcPatternDel (XFC_PATTERN_PTR (pattern), FC_CHARSET);
-
- name = FcNameUnparse (XFC_PATTERN_PTR (pattern));
+ {
+ FcPattern* temp = FcPatternDuplicate (XFC_PATTERN_PTR (pattern));
+ FcPatternDel (temp, FC_CHARSET);
+ name = FcNameUnparse (XFC_PATTERN_PTR (pattern));
+ FcPatternDestroy (temp);
+ }
result = build_fcapi_string (name);
xfree (name);
return result;
diff --git a/src/font-mgr.h b/src/font-mgr.h
--- a/src/font-mgr.h
+++ b/src/font-mgr.h
＠＠ -145,8 +145,10 ＠＠
do { \
DECLARE_EISTRING (eistrpxft_name); \
Extbyte *name; \
- FcPatternDel(pattern, FC_CHARSET); \
- name = (Extbyte *) FcNameUnparse (pattern); \
+ FcPattern* temp = FcPatternDuplicate (pattern); \
+ FcPatternDel (temp, FC_CHARSET); \
+ name = (Extbyte *) FcNameUnparse (temp); \
+ FcPatternDestroy (temp); \
eicpy_ext(eistrpxft_name, \
name ? name : "FONT WITH NULL NAME", \
Qfc_font_name_encoding); \
diff --git a/src/fontcolor-x.c b/src/fontcolor-x.c
--- a/src/fontcolor-x.c
+++ b/src/fontcolor-x.c
＠＠ -763,8 +763,12 ＠＠
"Xft font present but lacks pattern",
wrap_font_instance(f), Qfont, errb);
}
- FcPatternDel (pattern, FC_CHARSET); /* FcNameUnparse may choke */
- res = FcNameUnparse (pattern);
+ {
+ FcPattern* temp = FcPatternDuplicate (pattern);
+ FcPatternDel (temp, FC_CHARSET); /* FcNameUnparse may choke */
+ res = FcNameUnparse (temp);
+ FcPatternDestroy (temp);
+ }
if (res)
{
FONT_INSTANCE_TRUENAME (f) =
_______________________________________________
XEmacs-Patches mailing list
XEmacs-Patches(a)xemacs.org
http://lists.xemacs.org/mailman/listinfo/xemacs-patches
2:21 a.m.
APPROVE 21.5
That said, I don't like this because of the memory allocation implied,
and because I suspect my own patch was too intrusive. The crashes
seem to be only from the PRINT_XFT_PATTERN code (at least for me). I
don't understand why there are crashes there but not elsewhere yet.
The patch is OK for now, and I'll open an issue for it later, but I
wanted to raise the question here now so that others who have looked
at the code will be aware of it.
Steve
Michael Sperber writes:
This patch ...
2014-10-18 Stephen J. Turnbull <stephen(a)xemacs.org>
* fontcolor-xlike-inc.c (xft_find_charset_font):
* fontcolor-x.c (x_font_instance_truename):
* font-mgr.c (Ffc_name_unparse):
Protect Ffc_name_unparse from FcNameUnparse crash on charset property.
... caused some breakage for me. In particular, the test whether a
particular font implements a charset would always fail. The reason is
that fc patterns passed as arguments are modified in place. This patch
fixes that.
I'll push on Monday if nobody objects.
2014-10-25 Michael Sperber <mike(a)xemacs.org>
* fontcolor-x.c (x_font_instance_truename):
* font-mgr.c (Ffc_name_unparse):
* font-mgr.h (PRINT_XFT_PATTERN): Don't modify
fontconfig pattern passed in in place.
--
Regards,
Mike
diff --git a/src/font-mgr.c b/src/font-mgr.c
--- a/src/font-mgr.c
+++ b/src/font-mgr.c
＠＠ -266,9 +266,12 ＠＠
CHECK_FC_PATTERN (pattern);
/* #### Could use multiple values here to extract and return charset? */
- FcPatternDel (XFC_PATTERN_PTR (pattern), FC_CHARSET);
-
- name = FcNameUnparse (XFC_PATTERN_PTR (pattern));
+ {
+ FcPattern* temp = FcPatternDuplicate (XFC_PATTERN_PTR (pattern));
+ FcPatternDel (temp, FC_CHARSET);
+ name = FcNameUnparse (XFC_PATTERN_PTR (pattern));
+ FcPatternDestroy (temp);
+ }
result = build_fcapi_string (name);
xfree (name);
return result;
diff --git a/src/font-mgr.h b/src/font-mgr.h
--- a/src/font-mgr.h
+++ b/src/font-mgr.h
＠＠ -145,8 +145,10 ＠＠
do { \
DECLARE_EISTRING (eistrpxft_name); \
Extbyte *name; \
- FcPatternDel(pattern, FC_CHARSET); \
- name = (Extbyte *) FcNameUnparse (pattern); \
+ FcPattern* temp = FcPatternDuplicate (pattern); \
+ FcPatternDel (temp, FC_CHARSET); \
+ name = (Extbyte *) FcNameUnparse (temp); \
+ FcPatternDestroy (temp); \
eicpy_ext(eistrpxft_name, \
name ? name : "FONT WITH NULL NAME", \
Qfc_font_name_encoding); \
diff --git a/src/fontcolor-x.c b/src/fontcolor-x.c
--- a/src/fontcolor-x.c
+++ b/src/fontcolor-x.c
＠＠ -763,8 +763,12 ＠＠
"Xft font present but lacks pattern",
wrap_font_instance(f), Qfont, errb);
}
- FcPatternDel (pattern, FC_CHARSET); /* FcNameUnparse may choke */
- res = FcNameUnparse (pattern);
+ {
+ FcPattern* temp = FcPatternDuplicate (pattern);
+ FcPatternDel (temp, FC_CHARSET); /* FcNameUnparse may choke */
+ res = FcNameUnparse (temp);
+ FcPatternDestroy (temp);
+ }
if (res)
{
FONT_INSTANCE_TRUENAME (f) =
_______________________________________________
XEmacs-Patches mailing list
XEmacs-Patches(a)xemacs.org
http://lists.xemacs.org/mailman/listinfo/xemacs-patches
_______________________________________________
XEmacs-Patches mailing list
XEmacs-Patches(a)xemacs.org
http://lists.xemacs.org/mailman/listinfo/xemacs-patches
7:50 a.m.
New subject: [AC] Unbreak the umlauts
Thanks for the review - I've pushed the patch.
"Stephen J. Turnbull" <stephen(a)xemacs.org> writes:
That said, I don't like this because of the memory allocation
implied,
and because I suspect my own patch was too intrusive. The crashes
seem to be only from the PRINT_XFT_PATTERN code (at least for me). I
don't understand why there are crashes there but not elsewhere yet.
Yeah, it bothers me too - especially as fontconfig doesn't seem to pass
out-of-memory to the caller. On the other hand, there's plenty of that
allocation all over the fontconfig code.
--
Regards,
Mike
_______________________________________________
XEmacs-Patches mailing list
XEmacs-Patches(a)xemacs.org
http://lists.xemacs.org/mailman/listinfo/xemacs-patches
3:02 p.m.
Ar an cúigiú lá is fiche de mí Deireadh Fómhair, scríobh Michael Sperber:
diff --git a/src/font-mgr.c b/src/font-mgr.c
--- a/src/font-mgr.c
+++ b/src/font-mgr.c
＠＠ -266,9 +266,12 ＠＠
CHECK_FC_PATTERN (pattern);
/* #### Could use multiple values here to extract and return charset? */
- FcPatternDel (XFC_PATTERN_PTR (pattern), FC_CHARSET);
-
- name = FcNameUnparse (XFC_PATTERN_PTR (pattern));
+ {
+ FcPattern* temp = FcPatternDuplicate (XFC_PATTERN_PTR (pattern));
+ FcPatternDel (temp, FC_CHARSET);
+ name = FcNameUnparse (XFC_PATTERN_PTR (pattern));
Too late for the review, but this should be FcNameUnparse (temp), as it is in
the rest of the patch.
+ FcPatternDestroy (temp);
+ }
result = build_fcapi_string (name);
xfree (name);
return result;
--
‘Tramadol is further fed to cattle […] when working them […] (as draft
animals) so that the animals do not get tired quickly. …’
— Angewandte Chemie, Sept 2014, describing the social context of
(synthetic) tramadol having been found in Cameroon tree roots.
_______________________________________________
XEmacs-Patches mailing list
XEmacs-Patches(a)xemacs.org
http://lists.xemacs.org/mailman/listinfo/xemacs-patches
3608
days inactive
3633
days old
3 comments
3 participants
participants (3)
-
Aidan Kehoe -
Michael Sperber -
Stephen J. Turnbull