MSVC buildfarm members are all unhappy
... and the build logs don't show any particular reason for it.
What is wrong, and why isn't the buildfarm script capturing a
useful error message?
regards, tom lane
Tom Lane wrote:
... and the build logs don't show any particular reason for it.
What is wrong, and why isn't the buildfarm script capturing a
useful error message?
It's the format change in the makefile for foreign stuff. The line:
Could not match in foreign makefile
I don't know why it ends up where it does - when I run the build in my
VM, it ends up as the last line which makes it stand out properly.
I'll take a look at fixing the actual issue.
//Magnus
On Wed, Feb 25, 2009 at 4:40 PM, Tom Lane <tgl@sss.pgh.pa.us> wrote:
... and the build logs don't show any particular reason for it.
What is wrong, and why isn't the buildfarm script capturing a
useful error message?
Looks like this:
http://anoncvs.postgresql.org/cvsweb.cgi/pgsql/src/backend/foreign/Makefile?rev=1.2
broke Mkvcbuild.pm. The attached patch just comments out the offending
code in Mkvcbuild.pm - I assume Peter will be putting the FDW stuff
back in 8.5.
--
Dave Page
EnterpriseDB UK: http://www.enterprisedb.com
Attachments:
fdw.patchapplication/octet-stream; name=fdw.patchDownload
Index: src/tools/msvc/Mkvcbuild.pm
===================================================================
RCS file: /projects/cvsroot/pgsql/src/tools/msvc/Mkvcbuild.pm,v
retrieving revision 1.36
diff -u -r1.36 Mkvcbuild.pm
--- src/tools/msvc/Mkvcbuild.pm 21 Jan 2009 10:30:02 -0000 1.36
+++ src/tools/msvc/Mkvcbuild.pm 25 Feb 2009 17:09:51 -0000
@@ -312,14 +312,14 @@
$p->AddReference($postgres);
}
- $mf = Project::read_file('src\backend\foreign\Makefile');
- $mf =~ s{\\s*[\r\n]+}{}mg;
- $mf =~ m{FDW\s*=\s*(.*)$}m || die 'Could not match in foreign makefile' . "\n";
- foreach my $foreign (split /\s+/,$1)
- {
- my $proj = $solution->AddProject($foreign . '_fdw', 'dll', 'foreign', 'src\backend\foreign\\' . $foreign);
- $proj->AddReference($postgres);
- }
+ # $mf = Project::read_file('src\backend\foreign\Makefile');
+ # $mf =~ s{\\s*[\r\n]+}{}mg;
+ # $mf =~ m{FDW\s*=\s*(.*)$}m || die 'Could not match in foreign makefile' . "\n";
+ # foreach my $foreign (split /\s+/,$1)
+ # {
+ # my $proj = $solution->AddProject($foreign . '_fdw', 'dll', 'foreign', 'src\backend\foreign\\' . $foreign);
+ # $proj->AddReference($postgres);
+ # }
$mf = Project::read_file('src\bin\scripts\Makefile');
$mf =~ s{\\s*[\r\n]+}{}mg;
Dave Page wrote:
On Wed, Feb 25, 2009 at 4:40 PM, Tom Lane <tgl@sss.pgh.pa.us> wrote:
... and the build logs don't show any particular reason for it.
What is wrong, and why isn't the buildfarm script capturing a
useful error message?Looks like this:
http://anoncvs.postgresql.org/cvsweb.cgi/pgsql/src/backend/foreign/Makefile?rev=1.2
broke Mkvcbuild.pm. The attached patch just comments out the offending
code in Mkvcbuild.pm - I assume Peter will be putting the FDW stuff
back in 8.5.
Since we removed it from the general Makefiles, I suggest we actually
remove it from the Mkvcbuild.pm file as well. it's still there in the
history - just like the general Makefiles.
Thoughts?
//Magnus
Magnus Hagander <magnus@hagander.net> writes:
Since we removed it from the general Makefiles, I suggest we actually
remove it from the Mkvcbuild.pm file as well. it's still there in the
history - just like the general Makefiles.
+1. Comments are not a substitute for having CVS history ...
regards, tom lane
Tom Lane wrote:
Magnus Hagander <magnus@hagander.net> writes:
Since we removed it from the general Makefiles, I suggest we actually
remove it from the Mkvcbuild.pm file as well. it's still there in the
history - just like the general Makefiles.+1. Comments are not a substitute for having CVS history ...
I've applied a version that does this. Hopefully that'll bring the
buildfarm back into green.
//Magnus