Re: Flyspell is to be incorporated to dictionaries-common, closing some {x}emacs bugs
On Tue, Oct 04, 2005 at 08:46:21AM +0200, Romain Francoise wrote:
> Agustin Martin <agustin.martin@hispalinux.es> writes:
>
> > Just to let you know that I will soon upload a new dictionaries-common
> > package that will include a recent flyspell.el integrated into
> > the common dictionaries system. This should allow closing some emacs
> > and xemacs bugs,
>
> What about emacs-snapshot? I would prefer if you didn't shadow its
> version of flyspell.el as it's usually very much up-to-date (being a CVS
> version). What do you think?
I think is not only a matter of up-to-date versions, but also of
integration. Since emacs-snapshot is a very special case, you have to decide
whether it should have a consistent spellchecking behavior with the other
emacsen flavors, or on the contrary extensively test the original emacs code.
This affects not only flyspell.el, but also ispell.el, which is currently
overriden. The state of both is rather different, flyspell.el is actively
maintained upstream, but is also extensively changed in emacs-cvs, so the
concept of up-to-date depends on the selected branch. ispell.el has been in
stand-by for some years, and seems that upstream is still very busy. Here
the emacs-cvs code makes a real difference, with a lot of bug fixes and
new features.
I have been trying to follow both {xemacs,emacs}-{ispell,flyspell}.el CVS to
incorporate bug fixes and new features, and tried to contribute to the
emacs-cvs version when possible. But I did that in a rather cautious way,
and always having upstream as reference, so the shipped ispell.el does not
contain all the emacs-cvs changes, only those that fixed bugs.
I think we can try disabling emacs-snapshot integration into the common
system, so neither ispell.el or flyspell.el are overriden, and see if that
has any wrong side effect other than the lack of integration, for instance,
dicts not present in ispell-dictionary-alist will not have a way to say
that they are available and which parameters should be used. We might
reconsider this later if problems appear. We could also consider what is
needed for a minimal integration, but from the emacs-snapshot files.
What do you think?
--
Agustin
Reply to: