[Top][All Lists]

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: push_string_init calling check

From: Eric Blake
Subject: Re: push_string_init calling check
Date: Wed, 10 May 2006 16:06:37 +0000 (UTC)
User-agent: Loom/3.14 (http://gmane.org/)

Ilya N. Golubev <gin <at> mo.msk.ru> writes in Jan 2006:

> CVS branch: branch-1_4
> <recursive push_string with non-gnu cc> posted on Thu, 10 Nov 2005
> 18:08:17 +0300 (<027a4373626100-gin <at> mo.msk.ru>) describes an internal
> error that occurs in certain build configurations unconditionally once
> certain line of C code is executed, one with `push_string_init ()'
> passed to macro.  Batch oriented tests done by `make check' should
> readily detect that, but they do not, all of them are passed.  `gcov'
> shows that, when m4 built with gcc, that line is never executed by
> tests either.  Fixing that as follows.  (Certainly test file name is
> unimportant and may be easily changed as long as it matches
> `*[0-9][0-9].*'.)

Sorry for the delay, and thanks for the reminder to update the testsuite.  
Actually, tests are parsed out of m4.texinfo.  So this rough draft patch would 
create such a test.  I validated that I could make the m4 testsuite fail if I 
changed obstack.h to avoid the __GNUC__ optimization and reverted your patch 
from November, once this patch was applied to the documentation.

I will clean up the patch, and commit it to CVS once I have some more time; I 
will also forward port it to CVS head (since all the generated test files are 
currently checked into branch-1_4, my patch will have to regenerate them; I 
also got some spurious failures due to test numbers shifting around).

Index: doc/m4.texinfo
RCS file: /sources/m4/m4/doc/m4.texinfo,v
retrieving revision
diff -u -r1. m4.texinfo
--- doc/m4.texinfo      9 May 2006 04:07:21 -0000
+++ doc/m4.texinfo      10 May 2006 16:04:37 -0000
@@ -1815,6 +1815,16 @@
 @result{}Macro [foo].
 @end example
address@hidden Test the bug fixed in Nov 2005.
address@hidden example
address@hidden ignore
 Changing the quotes to the empty strings will effectively disable the
 quoting mechanism, leaving no way to quote text.

reply via email to

[Prev in Thread] Current Thread [Next in Thread]