On Wed, Jul 13, 2011 at 12:28:07PM +0200, Fabian Groffen wrote:
> On 13-07-2011 12:02:50 +0200, Stefan Manegold wrote:
> > On Wed, Jul 13, 2011 at 11:49:38AM +0200, Fabian Groffen wrote:
> > > On 13-07-2011 11:26:32 +0200, Stefan Manegold wrote:
> > > > > > Where exactly is this (still) a problem?
> > > > >
> > > > > on volund
> > > >
> > > > and how, i.e., which compiler, which configure options, which error
> > > > message?
> > > >
> > > > It works fine on all Fedora + gcc combinations I tested and also seems
> > > > to
> > > > work fine with nightly testing --- Apr2011 except Darwin & Solaris, the
> > > > latter now fixed; Aug2011 except Darwin, now fixed and work with
> > > > default...
> > >
> > > any compiler
> > >
> > > MCFLAGS -march=native -pipe -g -W -Wall -Wextra -O2
> > > MX_CFLAGS -Werror -Werror-implicit-function-declaration
> > > -Wpointer-arith -Wdeclaration-after-statement -Wp,-D_FORTIFY_SOURCE=2
> > >
> > > env CFLAGS='${MCFLAGS}' X_CFLAGS='${MX_CFLAGS}' ./configure
> > > --disable-dependency-tracking --disable-strict --disable-optimize
> > > --disable-debug
> >
> > I see (could not reproduce, yet, though ...).
> >
> > And what exactly is the error you get?
>
> it just dies after the warnings:
>
> Making MonetDB in
> /ufs/fabian/scratch/vtmp/build-Apr2011-volund.ins.cwi.nl/all
> MapiLib.pm.c: In function ‘SWIG_Perl_ConvertPtrAndOwn’:
> MapiLib.pm.c:1199:2: warning: value computed is not used
> MapiLib.pm.c: In function ‘SWIG_Perl_MakePtr’:
> MapiLib.pm.c:1226:7: warning: value computed is not used
> MapiLib.pm.c: In function ‘boot_MapiLib’:
> MapiLib.pm.c:6382:3: warning: unused variable ‘items’
>
> warnings treated as errors
Now I understand. thanks.
The problem is that you set X_CFLAGS yourself (in particular -Werror), while
using --disable-strict. Thus, our configure and Makefile code is "unware"
of this can does (can) not locally and selectively disable (via
GCC_SWIG_CFLAGS) (only) those warnings that we cannot do anything about in
SWIG-generated code.
(in this case, e.g., GCC_SWIG_CFLAGS='-Wno-unused-value -Wno-unused-variable')
Open question is, do we "give in" and accept that we disable all X_CFLAGS
for SWIG-related code (that includes our *.i origin and what swig generated
from that),
or do we rather aim at disabling only those checks that knowingly cause an
anavoidable problem with the actually SWIG-generated code, keeping all other
checks for our (and SWIG-generated) code in tact to monitor future code
changes?
Indeed, we should not sort this out in the Apr2011 branch, but only in the
Aug2011 branch, or even the default branch ...
Thanks, again!
Stefan
--
| Stefan.Manegold @ CWI.nl | DB Architectures (INS1) |
| http://CWI.nl/~manegold/ | Science Park 123 (L321) |
| Tel.: +31 (0)20 592-4212 | 1098 XG Amsterdam (NL) |
_______________________________________________
Checkin-list mailing list
[email protected]
http://mail.monetdb.org/mailman/listinfo/checkin-list
