Re: What holds the release (was: mouse-1-click-follows-link)

[Chong Yidong]

> The thing is, the trickiest-sounding items in FOR-RELEASE -- those
> under the "* FATAL  ERRORS" section -- also sound like some of the
> most important.  Releasing an Emacs that crashes in recent kernels
> isn't so nice...
>
> [Many of the other items seem not so important to me; I suppose if
> nobody can be found to fix them, many of those could just be dropped
> without noticeably decreasing the quality of the release.]

It may be better to be specific, so here are the contents of FOR-RELEASE,
as well as some of my opinions on them.  I haven't contributed as much to
Emacs as the others on this list, so my opinion about various bugs "not
worth blocking the release" may seem arrogant.  However, I'm pretty
frustrated, because helping to get the 22.1 release out the door seems
like a sisyphean effort -- more stuff keeps getting into FOR-RELEASE, and
the release that's coming "soon" keeps getting pushed back!

It's a vicious cycle -- people want to check in every last bugfix, because
the next release will be year away.  But the reason for the long release
cycles is that big bugfixes keep getting checked in, creating still more
bugs to fix!  In many projects, a "feature freeze" includes a moratorium
on fixes for fringe problems, or those that can't be fixed without major
destabilizing changes.  Several of the items in FOR-RELEASE seem to be of
this variety.

> Make VC-over-Tramp work where possible, or at least fail
> gracefully if something isn't supported over Tramp.
> To be done by Andre Spiegel <address@hidden>.

This has been sitting there for several months now.  Has it already been
done?  Is it a major inconvenience worth delaying 22.1 for?

> define-minor-mode should not put :require into defcustom.
> See msg from rms to emacs-devel on 21 Dec.

The relevant URL is

http://lists.gnu.org/archive/html/emacs-devel/2004-12/msg00732.html

I've tried working on this, with no result.  I couldn't find a clean way
to implement RMS's suggestion, and I'm not sure it even makes sense.  A
workaround for the particular problem originally reported by Stephen Stahl
is to add (require 'font-lock) to font-core.el and a ":require 'font-lock"
tag to the definition of global-font-lock-mode.  Maybe we should just do
that, and wait for 22.2 for whatever general solution is required.  (It
would be *nice* to get it into 22.1, but what it would *not* be nice to
delay 22.1 into 2006 just for that.)

> ** Update Speedbar.

The relevant URL is

http://lists.gnu.org/archive/html/emacs-devel/2005-05/msg00180.html

It sounds like this will be fixed soon.

> Enhance scroll-bar to handle tall line (similar to line-move).

This is a feature request, not a show-stopping bug.  In the real world,
99.9% of users don't use Emacs as an image viewer; they use a specialized
image viewing program.  Emacs is currently a mediocre image viewer; it
would be nice if it were a great image viewer, but not essential for the
release.

> Adapt mouse-sel-mode to mouse-1-click-follows-link.

I fixed this a couple weeks ago.  This entry should be removed.

> Make unexec handle memory mapping policy of the latest versions of
> Linux.

If I understand correctly, this problem only crops up on Red Hat systems
running ExecShield, which has been known to cause problems with other
programs -- not just Emacs.  Furthermore, this is a non-trivial bugfix, so
introducing it at this stage will probably create still more bugs, which
will delay the release yet again...

> Investigate reported crashes in compact_small_strings.

This seems to be a real bug.  Relevant URLs are:

http://lists.gnu.org/archive/html/emacs-pretest-bug/2004-10/msg00374.html
http://lists.gnu.org/archive/html/emacs-pretest-bug/2004-11/msg00143.html
http://lists.gnu.org/archive/html/bug-gnu-emacs/2005-04/msg00224.html

In the last of these messages, the reporter said he could get "reliable"
crashes, but somehow the discussion wasn't followed up.

> Investigate reported crashes related to using an invalid pointer
> from string_free_list.

I couldn't figure out which bug report this refers to.

> Ange-ftp should ignore irrelevant IPv6 errors:

I've tried contacting the bug reporter, but he has not responded to any of
my three messages over the last two months.  My DNS server also returns
IPV6 addresses, but I do not experience this problem.  I suspect it only
shows up for rare broken configurations that include a router and/or FTP
program that doesn't handle IPV6 properly.  It may not be worthwhile
delaying the release for this.

> Make GTK scrollbars behave like others w.r.t. overscrolling.

I don't experience any problem with GTK scrollbars.  It does not seem to
be a show-stopper.

> Avoid unbreakable loops in redisplay.

This is an "it would be nice" feature, not a show-stopper.  It would be
nice to have a safety feature to avoid running inappropriate display
properties, but AFAIK people aren't actually being affected by such bugs. 
This shouldn't block the release.

> Finish updating the Emacs Lisp manual.

Seems pretty much done.

> Update lispref/README.

What needs to be done here?

> Update the Emacs manual.

Seems done.

> Update AUTHORS.

Already done.

> Reorder NEWS entries.

Already done.

> Check the Emacs manual.

Already done (some chapters have only been checked by one guy, not two.)