This is the mail archive of the gdb-patches@sourceware.org mailing list for the GDB project.


Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]
Other format: [Raw text]

[patch 2/2] [mingw+non-GNU?] gnulib fnmatch for libiberty/ FNM_FILE_NAME bugs


Hi,

as described in the previous [patch 1/2] libiberty/fnmatch.c contains various
FNM_FILE_NAME bugs.  Currently FNM_FILE_NAME is not used anywhere by
gcc/gdb/binutils.

While these bugs can be attempted to be fixed (I did fix the '*' matching '/'
bug locally, it is not difficult, just there are too many bugs anyway, see
glibc 299a95b9f048679f1288512b0a6ab6ca16dd9d7c and
4f54cdb1dd4b8929231f42e22b79a0d5c7600cba for example):

(a) They do not affect glibc systems where bug-free system fnmatch gets used
    instead.

(b) gnulib already contains bug-free fnmatch implementation.

So I tried to import fnmatch into gdb/gnulib/ but it brings in 592KB of patch
as gnulib fnmatch brings in many other supportive gnulib functions.

On GNU/Linux fortunately nothing(*) gets built out of it.  Therefore
personally I am perfectly fine with importing this patch to trunk, it only
increases .tar.bz2 sources size.  But I expect that it may not be welcome, not
sure.

(*) almost - only 100 bytes of strnlen1

Still it seems to me as the most viable way how to fix set auto-load safe-path
wildcards for MinGW and possibly also non-GNU UNIX systems.  Any opinions?


Thanks,
Jan

Attachment: fnmatchgnu.patch.gz
Description: GNU Zip compressed data


Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]