Re: [ace-users] [autoconf] Crude script to compare autogeneratedconfig.h with regular one

0 views
Skip to first unread message

Johnny Willemsen

unread,
Aug 10, 2007, 12:15:29 PM8/10/07
to Thomas Girard, ace-...@cs.wustl.edu
Hi,

Thanks for the overview. Can you go through the list and see if the defines
autoconf doesn't set are really needed in general (they could be left over
after cleanup). If you think autoconf check for a certain feature, can you
report each macro seperately in bugzilla? Then we get a clean overview in
bugzilla what has to be done.

Regards,

Johnny Willemsen
Remedy IT
Postbus 101
2650 AC Berkel en Rodenrijs
The Netherlands
www.theaceorb.nl / www.remedy.nl

*** Integrated compile and test statistics see
http://scoreboard.theaceorb.nl ***
*** Commercial service and support for ACE/TAO/CIAO ***
*** See http://www.theaceorb.nl/en/support.html ***

> Hello,
>
> I've written the attached script to compare autogenerated
> config.h with
> regular one.
>
> Example:
> $ export ACE_ROOT=/path/to/ACE_wrappers
> $ cd $ACE_ROOT
> $ ./ace-cmp.sh ./objdir/ace/config.h ./ace/config-linux.h
> will compare autogenerated config from $ACE_ROOT/objdir/ace/config.h
> with the regular $ACE_ROOT/ace/config-linux.h
>
> This script is buggy and probably not portable - for instance it uses
> `cpp -dM'. I still have to find out why it does not reveal the
> ACE_HAS_POSIX_SEM_TIMEOUT difference pointed out last month.
>
> Anyway, on my box it shows 115 differences, and these will probably
> require some work.
>
> I'll be away for the week-end, hence the early post.
>
> Regards,
>
> Thomas
>

Thomas Girard

unread,
Aug 12, 2007, 1:31:58 PM8/12/07
to Johnny Willemsen, ace-...@cs.wustl.edu
Le vendredi 10 août 2007 à 18:15 +0200, Johnny Willemsen a écrit :
> Hi,
>
> Thanks for the overview. Can you go through the list and see if the defines
> autoconf doesn't set are really needed in general (they could be left over
> after cleanup). If you think autoconf check for a certain feature, can you
> report each macro seperately in bugzilla? Then we get a clean overview in
> bugzilla what has to be done.

Right, I will do this.

I have just found out why I was not getting the expected difference on
ACE_HAS_POSIX_SEM_TIMEOUT: ACE_HAS_LINUX_NPTL was *not* defined in the
regular $ACE_ROOT/ace/config-linux.h, hence all posix stuff was not
checked. It this intended?

Adding -DACE_HAS_LINUX_NPTL to the already fragile script gives the
attached diff I'll use to report bugs.

Regards,

Thomas

ace_5.5.10_linux_amd64_auto_to_regular.diff

Johnny Willemsen

unread,
Aug 13, 2007, 12:38:22 AM8/13/07
to Thomas Girard, ace-...@cs.wustl.edu
Hi,

> Right, I will do this.
>
> I have just found out why I was not getting the expected difference on
> ACE_HAS_POSIX_SEM_TIMEOUT: ACE_HAS_LINUX_NPTL was *not* defined in the
> regular $ACE_ROOT/ace/config-linux.h, hence all posix stuff was not
> checked. It this intended?

No, I don't think so. ACE_HAS_LINUX_NPTL seems something that could be
defined, but the include of config-posix.h shouldn't depend on that

> Adding -DACE_HAS_LINUX_NPTL to the already fragile script gives the
> attached diff I'll use to report bugs.

If you have things that need to be changed, let us know.

Johnny

Thomas Girard

unread,
Aug 13, 2007, 9:01:50 AM8/13/07
to ace-...@cs.wustl.edu
Le dimanche 12 août 2007 à 19:31 +0200, Thomas Girard a écrit :
> Adding -DACE_HAS_LINUX_NPTL to the already fragile script gives the
> attached diff I'll use to report bugs.

The script was not including ace/config-g++-common.h because it was
using cpp instead of g++. A modified script, fixing this issue and
defining ACE_HAS_LINUX_NPTL, is attached.

I'll try to improve this script when I have time so that:
* it does not need -DACE_HAS_LINUX_NPTL (not needed for ACE > 5.5.10)
* it works with other compilers

A new diff is attached as well. Lines starting with `-' are defined by
the autotools method while line starting with `+' are by the regular
method. There are a few false positive (e.g. the difference on gperf,
#define ACE_GPERF "gperf-ace", which is manually set for Debian
packages), but there are also interesting lines, like the difference on
ACE_DEFAULT_BASE_ADDR, yet to be explained.

Regards,

Thomas

ace-cmp.sh
ace_5.5.10_linux_amd64_auto_to_regular.diff
Reply all
Reply to author
Forward
0 new messages