On Mon, 2006-07-31 at 10:26 -0400, Derek Atkins wrote: > Ahh, looks like I was pointing to an old set of gconf schemas.. Thanks > to cstim for pointing this out to me.
We need to be more intelligent about adding new gconf keys. Had that new key been added as "hide_splash_screen" instead of "show_splash_screen", then the default value for unknown boolean keys (i.e. false) would have produced the same behavior after the patch as gnucash had before the patch. I'm not criticizing andi5. I should have caught this in my review, only I wasn't think in terms of the behavior across the transition from pre- to post- patch, only whether the patch was technically correct. This transition is something that should be considered any time a new gconf key is added. > I think the fact that gnucash can't (at runtime) point to gconf > schemas is going to bite us in the long run... No application can do this. Gconf is a client/server architecture where the keys are read/written by the gconfd process and passed via an ipc mechanism to gnucash. AFAICT, to do what you propose would require bypassing gconfd or just ditching gconf completely and reimplementing a private configuration storage mechanism. Also, remember that you're a developer who has multiple, sometimes incompatible versions of gnucash installed. Users with a single installed version of gnucash should never see this problem. David _______________________________________________ gnucash-devel mailing list [email protected] https://lists.gnucash.org/mailman/listinfo/gnucash-devel
