11:57 a.m.
Giacomo Boffi writes:
Mats Lidell writes:
> >>>>> Aidan Kehoe <kehoea(a)parhasard.net> writes:
>
> > My suspicions are that this is a compiler bug; no-one but Giacomo
> > has reproduced it, right?
>
> Do you mean bug in gcc 4.7.3?
no,
,----
| % gcc --version
| gcc (Debian 4.8.1-6) 4.8.1
| ...
| % CC=/usr/bin/gcc-4.7 ./configure $USUAL_OPTIONS
| ...
| % CC=/usr/bin/gcc-4.7 make beta
| ...
| % make check
| ...
| weak-tests.el: 140 of 140 tests successful (100%).
|
| Done
| make[1]: Leaving directory `/home/boffi/src/xemacs/src'
| %
`----
it seems it's either a compiler bug or a XEmacs bug triggered by the
new compiler --- I've just verified that a new version og gcc was made
available in the unstable repository, so in a few minutes I'll let you
know something more
not a few minutes, anyway...
% gcc --version
gcc (Debian 4.8.1-7) 4.8.1
...
I configured and built XEmacs with the above gcc and it still fails
when evaluating the function test-chars in mule-tests
bye
g
--
This message has been scanned for viruses and
dangerous content by MailScanner, and is
believed to be clean.
_______________________________________________
XEmacs-Beta mailing list
XEmacs-Beta(a)xemacs.org
http://lists.xemacs.org/mailman/listinfo/xemacs-beta
12:34 p.m.
New subject: [Bug: 21.5-b33] SERIOUS XEMACS BUG: NULL POINTER LRECORD
>>>> Giacomo Boffi <giacomo.boffi(a)polimi.it> writes:
I configured and built XEmacs with the above gcc and it still fails
when evaluating the function test-chars in mule-tests
I tried gcc 4.9.0 with not problems. 4.7.3 works fine to.
On the other hand I could not use the exact same configuration. To
many packages that I don't have installed. So I had to remove some
'--with...' to get a build.
I also use make check as a test. With 100% success on the mule test
that would indicate that I could not reproduce it right?
I also tried to use the manual steps with loading test-harness etc but
then I get an error that Assert isn't known. I guess you get the
crasch before that so that is also an indication I cant reproduces it.
Yours
--
%% Mats
_______________________________________________
XEmacs-Beta mailing list
XEmacs-Beta(a)xemacs.org
http://lists.xemacs.org/mailman/listinfo/xemacs-beta
1:41 p.m.
New subject: [Bug: 21.5-b33] SERIOUS XEMACS BUG: NULL POINTER LRECORD
Mats Lidell writes:
>>>>> Giacomo Boffi <giacomo.boffi(a)polimi.it>
writes:
> I configured and built XEmacs with the above gcc and it still fails
> when evaluating the function test-chars in mule-tests
I tried gcc 4.9.0 with not problems. 4.7.3 works fine to.
Yeah, I have a 4.8.1 build on Mac OS X now, tests run to completion
with no failures in mule-tests.
On the other hand I could not use the exact same configuration.
If the compiler is unbroken, this test should not be affected by build
options.
I also use make check as a test. With 100% success on the mule test
that would indicate that I could not reproduce it right?
Correct.
I also tried to use the manual steps with loading test-harness etc
but
then I get an error that Assert isn't known. I guess you get the
crasch before that so that is also an indication I cant reproduces it.
No, Assert gets defined very early in the process. No Assert, no
tests.
_______________________________________________
XEmacs-Beta mailing list
XEmacs-Beta(a)xemacs.org
http://lists.xemacs.org/mailman/listinfo/xemacs-beta
2:58 p.m.
New subject: [Bug: 21.5-b33] SERIOUS XEMACS BUG: NULL POINTER LRECORD
>>>> Stephen J Turnbull <stephen(a)xemacs.org> writes:
No, Assert gets defined very early in the process. No Assert, no
tests.
The procedure I was referring to was this below, stated earlier in
this thread:
----------------------------------------------------------------------
i started src/xemacs
i load-library test-harness
i find-file tests/automated/mule-tests.el
in the buffer with mule-tests
i required bytecomp
i eval-last-sexp on (defun test-chars ...)
i eval-last-sexp on (test-chars )
----------------------------------------------------------------------
As I get it the Assert macro is not defined just by loading
test-harness so when evaling (test-chars ...), in the context given
above, it is unknown.
Yours
--
%% Mats
_______________________________________________
XEmacs-Beta mailing list
XEmacs-Beta(a)xemacs.org
http://lists.xemacs.org/mailman/listinfo/xemacs-beta
5:41 p.m.
New subject: [Bug: 21.5-b33] SERIOUS XEMACS BUG: NULL POINTER LRECORD
Mats Lidell writes:
>>>>> Stephen J Turnbull <stephen(a)xemacs.org>
writes:
> No, Assert gets defined very early in the process. No Assert,
no
> tests.
The procedure I was referring to was this below, stated earlier in
this thread:
Yeah, Assert is a macro gets defined by the functions that run tests,
like test-emacs-test-file. Then when they're done, it gets undefined.
To find out about what functions you can use just by loading
test-harness, you'll have to read the file (test-harness.el), it's not
well documented.
There may be more information in the Internals manual.
But I'm pretty sure that this is something special to Giacomo's
platform. We just got another "this can't happen" report, so I'm
suspecting there's a bad patch floating around the distros. Dunno
whether it's to XEmacs or to GCC, but if that's what it is, I'd guess
it's to GCC.
Steve
_______________________________________________
XEmacs-Beta mailing list
XEmacs-Beta(a)xemacs.org
http://lists.xemacs.org/mailman/listinfo/xemacs-beta
10:49 p.m.
New subject: [Bug: 21.5-b33] SERIOUS XEMACS BUG: NULL POINTER LRECORD
>>>> Stephen J Turnbull <stephen(a)xemacs.org> writes:
But I'm pretty sure that this is something special to
Giacomo's
platform.
So it seems. Given that we can't recreate it there is little we can do
at the moment. If there comes up some new idea of what to test I'll be
happy to assist.
Yours
--
%% Mats
_______________________________________________
XEmacs-Beta mailing list
XEmacs-Beta(a)xemacs.org
http://lists.xemacs.org/mailman/listinfo/xemacs-beta
4141
days inactive
4141
days old
5 comments
3 participants
participants (3)
-
Giacomo Boffi -
Stephen J. Turnbull -
The XEmacs Package Smoketest