>>>> "Jan" == Jan Vroonhof
<vroonhof(a)math.ethz.ch> writes:
Jan> sperber(a)informatik.uni-tuebingen.de (Michael Sperber [Mr. Preprocessor]) writes:
> >> This looks wrong. Packages go into a package hierarchy,
and ~/.xemacs
> >> isn't one.
Jan> It is/was in 21.1.
Is no more in 21.2.
> >> The general concept of having a single package
hierarchy
> >> into which user packages go is broken by now.
Jan> Huh? Why?
There are *two* now under ~/.xemacs: xemacs-packages and
mule-packages.
> >> Actually, looking at
> >> package-get.el (for the first time), the entire logic associated with
> >> `package-get-user-package-location' is complete bogus.
> Especially the latter is totally sloppy and should be rewritten.
Jan> Please define 'sloppy'.
There's no clear concept of where the package index is really supposed
to go. `package-get-locate-index-file' calls `locate-data-file', but
the location where the package index is *stored* is not inferred from
`data-directory-list' at all. Moreover,
`package-get-locate-index-file', as a last resort, just returns
`package-get-base-filename' without checking for its existence.
> There's clearly a blatant bug here, but I'd rather leave
this eyesore
> in to remind people things are really broken here, rather than going
> for the quick fix.
Jan> Please define broken? It al does what it is supposed to (modulo "~")
Jan> 1. Get the package index from the best location possible.
Jan> 2. Get the file locally using the standard XEmacs data file
Jan> algorithms.
Jan> 3. If the two don't match offer to update the local file if we can
Jan> write to it.
This is all cool.
Jan> 4. If we cannot write to it, offer to write it to a location where the
Jan> user is likely to be able to write and where XEmacs will find it
Jan> again (i.e. the etc directory of the users package hierarchy).
But there's a strange mixture of genericity (`locate-data-file') and
hard-wired path construction (like (expand-file-name "etc/"
package-get-user-package-location) ) going on.
--
Cheers =8-} Chipsy
Friede, Völkerverständigung und überhaupt blabla