10:56 p.m.
Hi,
Trying to get all versions of XEmacs in the package smoketest to build
properly. Now the time has come to analyze 21.1 (mule and
no-mule).
Both fail in the same way se:
http://labb.contactor.se/~matsl/smoketest/logs/2005-02-16_20-01_build-all...
When I run 21.1 on another machine I get this error:
----------------------------------------------------------------------
make[2]: Entering directory `/src/xemacs/packages/xemacs-packages/ocaml'
/src/xemacs/bin/xemacs -no-autoloads -vanilla -batch -eval '(setq
stack-trace-on-error t load-always-display-messages t load-ignore-out-of-date-elc-files t
load-show-full-path-in-messages t)' -eval '(setq load-path (list (expand-file-name
"mule" lisp-directory) lisp-directory))' -l
/src/xemacs/packages/package-compile.el -- xemacs-base fsf-compat -- -l font-lock -l
overlay -l add-log -l imenu -l camldebug -l inf-caml -f batch-byte-compile
auto-autoloads.el
Loading /src/xemacs/lib/xemacs-21.1.14/lisp/auto-autoloads...
Loading /src/xemacs/packages/xemacs-packages/fsf-compat/auto-autoloads...
Loading /src/xemacs/packages/xemacs-packages/xemacs-base/auto-autoloads...
No such face: font-lock-constant-face
xemacs exiting.
make[2]: *** [auto-autoloads.elc] Error 255
make[2]: Leaving directory `/src/xemacs/packages/xemacs-packages/ocaml'
make[1]: *** [ocaml/bytecompile.target] Error 2
make[1]: Leaving directory `/src/xemacs/packages/xemacs-packages'
make: *** [xemacs-packages/bytecompile.target] Error 2
----------------------------------------------------------------------
I think that the smoketest build gets the same error but fails to
report it in clear text due to that it hangs for some reason. (I have
a script that periodically checks for hanging xemacs processes and
kills them to avoid the whole smoketest to stop.) (Could the ^G, bell,
cause some problem when the process is run by cron!?)
Anyway the main issue here is that xemacs 21.1 gets a problem with the
missing font-lock-constant-face when compiling ocaml.
Yours
--
%% Mats
2:06 a.m.
>>>> "Mats" == Mats Lidell
<matsl(a)xemacs.org> writes:
Mats> Anyway the main issue here is that xemacs 21.1 gets a
Mats> problem with the missing font-lock-constant-face when
Mats> compiling ocaml.
Yup. There was no such standard face at the time.
I'm not sure what should be done about that. I suppose we could wrap
the offending code in a version test in our package, which would
effectively mark the backward-compatibility problem.
--
Institute of Policy and Planning Sciences http://turnbull.sk.tsukuba.ac.jp
University of Tsukuba Tennodai 1-1-1 Tsukuba 305-8573 JAPAN
Ask not how you can "do" free software business;
ask what your business can "do for" free software.
3:23 a.m.
* Stephen J Turnbull <stephen(a)xemacs.org> writes:
>>>>> "Mats" == Mats Lidell
<matsl(a)xemacs.org> writes:
Mats> Anyway the main issue here is that
xemacs 21.1 gets a
Mats> problem with the missing font-lock-constant-face when
Mats> compiling ocaml.
Yup. There was no such standard face at the time.
I'm not sure what should be done about that. I suppose we could
wrap
the offending code in a version test in our package, which would
effectively mark the backward-compatibility problem.
I thought packages no longer were required to support XEmacs 21.1 and
older?
I suggest removing the smoketests for 21.1.
--
|---<Steve Youngs>---------------<GnuPG KeyID: A94B3003>---|
| In space, |
| No one can hear you rip a stinky |
|------------------------------------<steve(a)sxemacs.org>---|
6 a.m.
>>>> "SY" == Steve Youngs
<steve(a)sxemacs.org> writes:
SY> I thought packages no longer were required to support XEmacs
SY> 21.1 and older?
Required, no. Will anybody do anything about it? Probably not. But
we may as well know about it, IMO.
SY> I suggest removing the smoketests for 21.1.
That's up to Mats. But I would rather not, if we can find a way to
keep them that's not burdensome for him.
--
Institute of Policy and Planning Sciences http://turnbull.sk.tsukuba.ac.jp
University of Tsukuba Tennodai 1-1-1 Tsukuba 305-8573 JAPAN
Ask not how you can "do" free software business;
ask what your business can "do for" free software.
11:19 a.m.
Stephen J. Turnbull wrote:
>>>>> "SY" == Steve Youngs
<steve(a)sxemacs.org> writes:
SY> I suggest removing the smoketests for 21.1.
That's up to Mats. But I would rather not, if we can find a way to
keep them that's not burdensome for him.
What is the rationale in building the packages with 21.1 if the package maintainers
isn't supposed to fix the problems that the build finds? Knowledge about which
packages
failes I guess but is it worth the trouble!?
Yours
--
%% Mats
12:28 p.m.
>>>> "Mats" == Mats Lidell
<matsl(a)xemacs.org> writes:
Mats> What is the rationale in building the packages with 21.1 if
Mats> the package maintainers isn't supposed to fix the problems
Mats> that the build finds?
I would hope the package maintainers would agree to fix such problems;
we just won't twist their arms any more.
--
Institute of Policy and Planning Sciences http://turnbull.sk.tsukuba.ac.jp
University of Tsukuba Tennodai 1-1-1 Tsukuba 305-8573 JAPAN
Ask not how you can "do" free software business;
ask what your business can "do for" free software.
2:25 a.m.
* Mats Lidell <matsl(a)xemacs.org> writes:
Stephen J. Turnbull wrote:
>
> I would hope the package maintainers would agree to fix such problems;
> we just won't twist their arms any more.
>
Considering the current problem, with the missing
font-lock-constant-face, is that possible to fix in some decent
way?
In the .el in question...
(when (and (featurep 'xemacs)
(eq emacs-major-version 21)
(eq emacs-minor-version 1))
(make-face 'font-lock-constant-face))
...would probably do it.
--
|---<Steve Youngs>---------------<GnuPG KeyID: A94B3003>---|
| In space, |
| No one can hear you rip a stinky |
|------------------------------------<steve(a)sxemacs.org>---|
6:45 a.m.
On Fri, 2005-02-18 at 12:25 +1000, Steve Youngs wrote:
* Mats Lidell <matsl(a)xemacs.org> writes:
> Considering the current problem, with the missing
> font-lock-constant-face, is that possible to fix in some decent
> way?
In the .el in question...
(when (and (featurep 'xemacs)
(eq emacs-major-version 21)
(eq emacs-minor-version 1))
(make-face 'font-lock-constant-face))
...would probably do it.
Can't the definedness of font-lock-constant-face be tested instead of
doing version checks?
font-lock-constant-face was added in 21.4.9 and 21.5.9.
9:01 a.m.
* Ville Skytt <Ville> writes:
Can't the definedness of font-lock-constant-face be tested
instead
of doing version checks?
Actually, simply doing (make-face 'font-lock-constant-face) without
any checks of any kind will do. `make-face' won't alter an already
existing face...
,----[ C-h f make-face RET ]
| `make-face' is a built-in function
| (make-face NAME &optional DOC-STRING TEMPORARY)
|
| Documentation:
| Define a new face with name NAME (a symbol), described by DOC-STRING.
| You can modify the font, color, etc. of a face with the set-face-* functions.
| If the face already exists, it is unmodified.
| If TEMPORARY is non-nil, this face will cease to exist if not in use.
`----
--
|---<Steve Youngs>---------------<GnuPG KeyID: A94B3003>---|
| In space, |
| No one can hear you rip a stinky |
|------------------------------------<steve(a)sxemacs.org>---|
noon
Steve Youngs wrote:
Actually, simply doing (make-face 'font-lock-constant-face)
without
any checks of any kind will do. `make-face' won't alter an already
existing face...
So for keeping 21.1 alive we could add such a code to all packages that needs it.
This seems less ideal so, alternatively, could that be placed in some generic forward
compatibility place so that it is only needed to add it once?
And most important will the code not only compile but work as expected as well? If
compile time errors only turn into run time errors not much is gained.
Yours
--
%% Mats
9:49 p.m.
>>>> Stephen wrote:
>>>> "SY" == Steve Youngs
<steve(a)sxemacs.org> writes:
SY> I suggest removing the smoketests for 21.1.
Stephen> That's up to Mats. But I would rather not, if we can find a
Stephen> way to keep them that's not burdensome for him.
The only way is to fix the problems so the builds get green
again. Since there seems little interest in fixing 21.1 build problems
I also suggest dropping the smoketest for 21.1. Constantly building
just to find the same problem over and over again doesn't seems to do
any good. It just makes XEmacs look bad with all these red builds
never getting fixed.
Yours
--
%% Mats
12:44 a.m.
>>>> "Mats" == Mats Lidell
<matsl(a)xemacs.org> writes:
Mats> Since there seems little interest in fixing 21.1 build
Mats> problems I also suggest dropping the smoketest for 21.1.
Your forward-compatibility package suggestion should work, and that
kind of thing has been suggested many times. We _will_ need it for
21.4 once we release 22.0, because things have really changed quite a
bit in 21.5 already.
But go ahead and drop the 21.1 smoketest. As you say, nobody seems
interested in fixing problems.
--
Institute of Policy and Planning Sciences http://turnbull.sk.tsukuba.ac.jp
University of Tsukuba Tennodai 1-1-1 Tsukuba 305-8573 JAPAN
Ask not how you can "do" free software business;
ask what your business can "do for" free software.
6987
days inactive
7010
days old
12 comments
4 participants
participants (4)
-
Stephen J. Turnbull -
Steve Youngs -
The XEmacs Package Smoketest -
Ville Skyttä