Re: patches in 0.99.19-208 (was Re: slang/slrn bug with CTRL-G / CTRL-Z combo)

["John E. Davis" <davis@xxxxxxxxxxxxx>]

On Thu, 19 Nov 2009 13:45:35 +0000 (UTC), Jörg Sommer <joerg@xxxxxxxxxxxx>
wrote:
> Now, I've some new problems:
>
> • make distclean doesn't remove log files of failed tests. The test
>   test_cmode failed and after running make distclean the file
>   src/test/test_cmode.log still exists. Distclean should remove
>   everything.
>
> • With the commit for the changes 205–207 you've changed the test data
>   for test_cmode, but this makes the test fail.

These have been fixed and will appear in the repository soon.

> And what's your opinion to the following patches? For the first to third
> patch please remember this mail [1]. I can't recall what the result of
> the discussion was, but without some tweaking [2] I get these warnings:
>
> dpkg-shlibdeps: warning: dependency on libdl.so.2 could be avoided if "debian/jed/usr/bin/jed" were not uselessly linked against it (they use none of its symbols).
> dpkg-shlibdeps: warning: dependency on libm.so.6 could be avoided if "debian/jed/usr/bin/jed" were not uselessly linked against it (they use none of its symbols).
> dpkg-shlibdeps: warning: dependency on libdl.so.2 could be avoided if "debian/xjed/usr/bin/xjed" were not uselessly linked against it (they use none of its symbols).
> dpkg-shlibdeps: warning: dependency on libm.so.6 could be avoided if "debian/xjed/usr/bin/xjed" were not uselessly linked against it (they use none of its symbols).
> dpkg-shlibdeps: warning: dependency on libgpm.so.2 could be avoided if "debian/xjed/usr/bin/xjed" were not uselessly linked against it (they use none of its symbols).

It is my understanding that not all systems have dynamic linkers that
can automatically link to dependent libraries.  Such systems require
the libraries to be explicitly linked.  

[...]
> Insert DYNAMIC_LINK_FLAGS directly into LDFLAGS
>
> There's no need to carry these flags in a separate variable.

I regard this as a cosmetic change.  Also, for robustness I tend to
share aclocal.m4 between a number of projects, and some need that
variable.

> Replace check for dlopen by check for SLang_init_import

This one is related to the library dependency issue mentioned above.

> Remove linker option -lm
>
> The library math is not used by Jed. So we don't need to link with it.

 ditto.

> http://git.debian.org/?p=pkg-jed/jed.git;a=commitdiff;h=8d62d0f896be1d03cb623d66623e057faca04091
> Author: Jörg Sommer <joerg@xxxxxxxxxxxx>
> Date:   Sun, 24 Aug 2008 17:41:40 +0200
>
> Make return type of jed_malloc0() void*

This is something that I intend to address for slang-3, where I will
change 'char *SLmalloc' to 'void *SLmalloc'.  I will not be able to do
it before then because the API must be backwards compatible.

> The cast from char* to any other pointer type might increases required
> alignment of target type. On some architectures (like alpha) this
> triggers an error message at runtime, if the access of a pointer is
> unaligned:

This will never happen in jed because jed_malloc0 is a wrapper around
SLmalloc, which in turn wraps malloc.  And malloc returns pointers
with the proper alignment.

I will add the rest of the patches.

I hope to release 0.99.19 at the end of the month.

Thanks,
--John

--------------------------
To unsubscribe send email to <jed-users-request@xxxxxxxxxxx> with
the word "unsubscribe" in the message body.
Need help? Email <jed-users-owner@xxxxxxxxxxx>.