HAL 0.1 release
lah at micropp.se
Thu Oct 2 10:26:04 EEST 2003
Havoc Pennington wrote:
>On Wed, 2003-10-01 at 13:39, David Zeuthen wrote:
>>Anyone else got a take on python vs. XML?
>XML. You want this to be data, not code. Mixing the two causes problems.
>Data is much more well-defined.
>Also I agree the python dependency is undesirable. Really ought to go
>with the zero deps here (especially for the client library; a daemon may
>be able to get away with something like GLib).
Have not played with glib for a long time and I don't know how well it
is spitt up and how much internal dependencis it have. But if the need
is for only a subset of glib, that don't depend on the rest of glib -
staticly linking to the exactly needed object files might be a solution.
That mean no dependencys on glib being instald or witch version of glib
being installd. It should also put all desktops on equal standing in
fotprint. But ther *might* still be some kind of 'political' dependence :-/
Off corse, it is a small fotprint cost compered with dynamicly linking
glib (if glib is used anyway by other aplikations), but not realy any
cost compered with reinventing the same stuff - and it would increese
robustness (compered with both alternative)!
Probably best way to do it is to include a snapshot of neded source
files from glib in the cvs/tarball and bould the neded objectfiles
during make. But still having the advantage of a upstream to send
bugfixes/enhansment to and a, by updating those files from time to time,
a upstream to resive others bugfixes/enhansment from. Less duplikation
of effort simply put.
More information about the xdg