Synopsis - Issue Tracker

Synopsis 0.8 fails to build with gcc 4.0.  Partial build log attached.

Oh, that's too bad ! The fix looks trivial, though
(a simple forward declaration).
I couldn't reproduce the error, though, using
a gcc 4.0.0 installation. What version did you
use ?

I'll check in a fix, and probably spin a bugfix
release (0.8.1) shortly.

Thanks for the report !

> Oh, that's too bad ! The fix looks trivial,
> though (a simple forward declaration).
> I couldn't reproduce the error, though, using
> a gcc 4.0.0 installation. What version did you
> use ?

gcc 4.0.1, with some CVS fixes.

> I'll check in a fix, and probably spin a bugfix
> release (0.8.1) shortly.

Thanks for the fast response!  It might be a good
idea for me to test the build first to make sure
there are no more errors.

Matej Vela via synopsis issue tracker wrote:

> Thanks for the fast response!  It might be a good
> idea for me to test the build first to make sure
> there are no more errors.

Fine. Could you please apply the attached patch to
the src/Synopsis/Python/Object.hh file to test the fix ?

I may as well compile a newer gcc version to test it
myself, but meanwhile...

Thanks,
		Stefan

> Could you please apply the attached patch to
> the src/Synopsis/Python/Object.hh file to test
> the fix ?

The build completed (though the patch needs to be
reversed and stripped of comments).

Unfortunately, "setup.py test" gives 70%
failures.  How do we debug this?

Thanks!

Matej

Matej Vela via synopsis issue tracker wrote:

> The build completed (though the patch needs to be
> reversed and stripped of comments).

Good. I installed gcc 4.0.1 here and tested the patch
as well. I'v now set up a 'Synopsis_0_8' branch that
contains this fix as well as a number of backported
modifications from the trunk. I'd like to spin a
0.8.1 bugfix release in the near future.

> Unfortunately, "setup.py test" gives 70%
> failures.  How do we debug this?

First of all, these failures are (mostly) known and
are used to measure progress as well as regressions
while I implement full support for ISO C++. Luckily,
most of the tested features aren't used when synopsis
runs as a documentation tool. They will be used as
I move forward transforming it into a true code
introspection tool.

As a reference you may compare your test results
to http://synopsis.fresco.org/tests/, which should
always provide the 'expected' test outcome.

--

Stefan

PS: I note that you have a debian account. Would you
     be interested in helping me to generate debian
     packages ? There is some preliminary support in
     the build system. Try 'python setup.py bdist_dpkg'...

> Good. I installed gcc 4.0.1 here and tested the
> patch as well. I'v now set up a 'Synopsis_0_8'
> branch that contains this fix as well as a
> number of backported modifications from the
> trunk. I'd like to spin a 0.8.1 bugfix release
> in the near future.

Great!

> As a reference you may compare your test results
> to http://synopsis.fresco.org/tests/, which
> should always provide the 'expected' test
> outcome.

I see a number of extra failures, but that may be
due to bug fixes in the trunk.  Anyway, I'm not
worried now that you have gcc 4.0.1.

> PS: I note that you have a debian account. Would
>     you be interested in helping me to generate
>     debian packages ? There is some preliminary
>     support in the build system. Try 'python
>     setup.py bdist_dpkg'...

I'm short on time right now, but I'll take a look
at it during the weekend.  The current Debian
package doesn't have a maintainer, but I'm pretty
sure I can update it to 0.8.1 once it's released.

Thanks!

Matej

What's the status for this ? I think both, stable
(branches/Synopsis_0_8/) as well as devel (trunk/)
are patched for this. Can you please confirm ?

Here's what I'm getting with gcc 4.0.3 for both
branches/Synopsis_0_8/ and trunk/...

Seems the log didn't make it, here it is.

I couldn't reproduce the error with a freshly installed gcc 4.0.2
Anyways, I checked in a fix into both branches. Can you please
try again ?

The trunk builds fine, but branches/Synopsis_0_8
fails with the following error:

make[2]: Entering directory `/tmp/Synopsis/build/ctemp.linux-i686/src/Synopsis/gc/doc'
make[2]: *** No rule to make target `README.darwin', needed by `all-am'.  Stop.
make[2]: Leaving directory `/tmp/Synopsis/build/ctemp.linux-i686/src/Synopsis/gc/doc'
make[1]: *** [all-recursive] Error 1

A bug in the Makefile?

Matej Vela via synopsis issue tracker wrote:
> Matej Vela <vela@debian.org> added the comment:
> 
> The trunk builds fine, but branches/Synopsis_0_8
> fails with the following error:
> 
> make[2]: Entering directory `/tmp/Synopsis/build/ctemp.linux-i686/src/Synopsis/gc/doc'
> make[2]: *** No rule to make target `README.darwin', needed by `all-am'.  Stop.
> make[2]: Leaving directory `/tmp/Synopsis/build/ctemp.linux-i686/src/Synopsis/gc/doc'
> make[1]: *** [all-recursive] Error 1

I have never seen this error before ! What platform are you on ?
Did you do a clean build or an incremental build ? (I recently upgraded the
gc sources to version 6.6, and in doing so removed the '.deps' directory
that used to be in the repository.)

A quick 'grep' shows the only mention of 'README.darwin' in Makefile.direct,
which isn't used (itself only mentioned as part of the EXTRA_DIST variable).

Can you try to figure out why make picks up the 'README.darwin' target at all ?
(The file is in the source directory, but we are building in a separate build dir.)

> I have never seen this error before ! What platform are you on ?

Debian unstable.  My guess is that make 3.81beta4
is the culprit -- it seems to detect a number of
problems that were ignored by earlier versions.

> Did you do a clean build or an incremental build ? (I recently upgraded the
> gc sources to version 6.6, and in doing so removed the '.deps' directory
> that used to be in the repository.)

It was a clean build.

> A quick 'grep' shows the only mention of 'README.darwin' in Makefile.direct,
> which isn't used (itself only mentioned as part of the EXTRA_DIST variable).
>
> Can you try to figure out why make picks up the 'README.darwin' target at all ?
> (The file is in the source directory, but we are building in a separate build dir.)

src/Synopsis/gc/doc/Makefile.am lists
README.darwin in $(dist_pkgdata_DATA).  In the
generated Makefile, $(dist_pkgdata_DATA) is
included in $(DATA), which is a requisite for the
all-am target; thus the error.

All in all, the following files appear to be
missing from branches/Synopsis_0_8_0:

  Synopsis/src/Synopsis/gc/darwin_stop_world.c
  Synopsis/src/Synopsis/gc/doc/README.darwin
  Synopsis/src/Synopsis/gc/doc/simple_example.html
  Synopsis/src/Synopsis/gc/include/private/darwin_semaphore.h
  Synopsis/src/Synopsis/gc/include/private/darwin_stop_world.h
  Synopsis/src/Synopsis/gc/include/private/pthread_stop_world.h
  Synopsis/src/Synopsis/gc/include/private/pthread_support.h
  Synopsis/src/Synopsis/gc/pthread_stop_world.c
  Synopsis/src/Synopsis/gc/pthread_support.c

Once they are copied from the trunk, the build
works as expected.

Matej Vela via synopsis issue tracker wrote:

> All in all, the following files appear to be
> missing from branches/Synopsis_0_8_0:

Doh ! I was missing a whole lot of files when upgrading to gc 6.6 !
They are all checked in now. Please confirm that it works now.

> Doh ! I was missing a whole lot of files when upgrading to gc 6.6 !
> They are all checked in now. Please confirm that it works now.

Ayup, builds fine.

Thanks!

Matej

Great, lets close this bug then.
As I'd like to plan for the next release, which will
still be based on the stable (i.e. Synopsis_0_8) 
branch, could you please look into any debian
packaging issues, and if there are any, open a new
bug ?
There are still some features I'd like to work 
out, but it would be nice if 0.9 could get out in
early April.
Many thanks !