Hi Marc,
Change looks fine and the test passes after applying the patch.
OK to push.
Thanks for fixing the failing unit tests so quickly.
Best regards,
Olav
Marc Alff wrote:
> #At file:///home/malff/BZR_TREE/mysql-next-mr-bugfixing-51741/ based on
> revid:alik@stripped
>
> 3122 Marc Alff 2010-03-04
> Bug#51741 Unit test pfs-t failing in mysql-next-mr-bugfixing
>
> The root cause of the failure is that when
> Bug#51447 performance schema evil twin files
> was fixed, instrumented file names got normalized.
>
> The pfs-t unit test depends on this file normalization,
> but it was not updated.
>
> This fix aligns pfs-t.cc lookup_file_by_name()
> with the logic in pfs_instr.cc find_or_create_file().
>
> modified:
> storage/perfschema/unittest/pfs-t.cc
> === modified file 'storage/perfschema/unittest/pfs-t.cc'
> --- a/storage/perfschema/unittest/pfs-t.cc 2010-01-12 01:47:27 +0000
> +++ b/storage/perfschema/unittest/pfs-t.cc 2010-03-05 01:10:27 +0000
> @@ -37,14 +37,25 @@ PFS_file* lookup_file_by_name(const char
> uint i;
> PFS_file *pfs;
> uint len= strlen(name);
> + size_t dirlen;
> + const char *filename;
> + uint filename_length;;
>
> for (i= 0; i < file_max; i++)
> {
> pfs= & file_array[i];
> if (pfs->m_lock.is_populated())
> {
> - if ((len == pfs->m_filename_length) &&
> - (strncmp(name, pfs->m_filename, pfs->m_filename_length) == 0))
> + /*
> + When a file "foo" is instrumented, the name is normalized
> + to "/path/to/current/directory/foo", so we remove the
> + directory name here to find it back.
> + */
> + dirlen= dirname_length(pfs->m_filename);
> + filename= pfs->m_filename + dirlen;
> + filename_length= pfs->m_filename_length - dirlen;
> + if ((len == filename_length) &&
> + (strncmp(name, filename, filename_length) == 0))
> return pfs;
> }
> }
>
>
> ------------------------------------------------------------------------
>
>
