From: Joerg Bruehe
Date: February 10 2011 9:48pm
Subject: Re: bzr commit into mysql-5.5-bugteam branch (joerg:3212) Bug#42969
List-Archive: http://lists.mysql.com/commits/131089
Message-Id: <4D545D1A.6020603@oracle.com>
MIME-Version: 1.0
Content-Type: text/plain; charset=ISO-8859-15
Content-Transfer-Encoding: quoted-printable

Hi Guilhem!


Guilhem Bichot wrote:
> Hello,
>=20
> Joerg Bruehe a =E9crit, Le 07.02.2011 12:50:
>> #At file:///MySQL/REPO/V55/bug42969-5.5/ based on
>> revid:joerg@stripped
>>
>>  3212 Joerg Bruehe    2011-02-07
>>       Bug #42969      Please add a MANIFEST to each build
>>             First attempt at a test that will verify the existence
>>       of both "INFO_SRC" and "INFO:_BIN", as well as (some of)
>>       their expected (invariant) contents.
>>      =20
>=20
>> +open(I_SRC,"<","$dir_docs/INFO_SRC") or print "Cannot open 'INFO_SRC'=

>> in '$dir_docs'\n";
>> +while(defined ($line =3D <I_SRC>)) {
>> +  if ($line =3D~ m|^MySQL source \d\.\d\.\d+|) {$found_version =3D "F=
ound
>> MySQL version number";}
>> +  if ($line =3D~ m|^revision-id: |) {$found_revision =3D "Found BZR
>> revision id";}
>=20
> Can we make the pattern above more specic, to verify that in the
> revision-id line we have '@' and '-2', as in
> joerg@stripped
> ?
> The interest is making sure that it's really a bzr revision-id which th=
e
>  cmake commands put in INFO_SRC, and not some wrong data.

Sure, that is no problem. I will do that.

>=20
>=20
>> +}
>> +close I_SRC;
>> +print "INFO_SRC: $found_version / $found_revision\n";
>> +$found_compiler =3D "No line 'Compiler flags'";
>> +$found_features =3D "No line 'Feature flags'";
>> +open(I_BIN,"<","$dir_docs/INFO_BIN") or print "Cannot open 'INFO_BIN'=

>> in '$dir_docs'\n";
>> +while(defined ($line =3D <I_BIN>)) {
>> +  if ($line =3D~ m| Compiler flags used |) {$found_compiler =3D "Foun=
d
>> 'Compiler flags' line";}
>=20
> here we don't search for ':'

Correct - the text doesn't have it:
1) That is my first version, it fails on Windows, because there I cannot
   output compiler *flags*, rather I output "Generator used".
2) The complete header line about "flags" (on Unix) is
      =3D=3D=3D=3D=3D Compiler flags used (from the 'sql/' subdirectory):=
 =3D=3D=3D=3D=3D
   because some directories are using slightly different flags, so you
   see there is no colon immediately after "used".
   On Windows, it is
      =3D=3D=3D=3D=3D Compiler / generator used: =3D=3D=3D=3D=3D

>=20
>> +  if ($line =3D~ m| Feature flags used:|) {$found_features =3D "Found=

>> 'Feature flags' line";}
>=20
> here we do?

Yes, as explained above (feature flags apply to the whole build).

The current check on "INFO_BIN" is

   while(defined ($line =3D <I_BIN>)) {
     # "generator" on Windows, "flags" on Unix:
     if (($line =3D~ m| Compiler / generator used: |) ||
         ($line =3D~ m| Compiler flags used |))   {$found_compiler =3D
"Found 'Compiler ... used' line";}
     if  ($line =3D~ m| Feature flags used:|)     {$found_features =3D
"Found 'Feature flags' line";}
   }

So I don't mention "flags" in the success message any more.

>=20
> The rest looks quite good.

Thanks!
I trust checking for two header lines per file is enough, and the
messages should clearly tell the problem in case of a failing test.

By now, I fixed the Windows issue, and I also fixed path name build
(RPMs use a different layout, even different for SuSE and RedHat).
I will check the results of the last platforms tomorrow (some are still
missing), but what I got till now already covers all different types of
platforms.

I expect to provide the final (combined) changeset tomorrow.


--=20
Joerg Bruehe,  MySQL Build Team,  joerg.bruehe@stripped
               (+49 30) 417 01 487
ORACLE Deutschland B.V. & Co. KG,   Komturstrasse 18a,   D-12099 Berlin
Geschaeftsfuehrer: Juergen Kunz, Marcel v.d. Molen, Alexander v.d. Ven
Amtsgericht Muenchen: HRA 95603