From: Mattern Thibaut<tma...@no...> - 2004-04-09 17:01:27
|
Citation de Daniel Caujolle-Bert <seg...@cl...>: > -----BEGIN PGP SIGNED MESSAGE----- > Hash: SHA1 > > Hi, > > Le vendredi 09 Avril 2004 16:01, Bastien Nocera a > écrit : > > On Fri, 2004-04-09 at 16:26, Daniel Caujolle-Bert > wrote: > > > -----BEGIN PGP SIGNED MESSAGE----- > > > Hash: SHA1 > > > > > > Hi, > > > > > > Le vendredi 09 Avril 2004 13:56, Miguel Freitas a > écrit : > > > > On Thu, 2004-04-08 at 18:47, Siggi Langauf wrote: > > > > > On Thu, 8 Apr 2004, Mattern Thibaut wrote: > > > > > > > > > > [...] > > > > > > > > > > > copy domain + "/" + user -> user param > > > > > > > > > > [...] > > > > > > > > > > > split returned user into domain, user using > "/" as separator > > > > > > > > > > That doesn't work if the user name contains "/". > > > > > > > > > > Additionally, it's just ugly. > > > > > > > > I agree with Siggi. > > > > > > > > I vote for heavily borrowing ideas from gnome > folks, and not > > > > reinventing > > > > > > Well, the thing i really want to avoid here is such > Gnome centric stuff. > > > IMHO we need to implement a generic (read universal) > feature, not a > > > specific one. > > > > That also means that xine wouldn't need to be modified > to handle a > > possible native smb plugin. > > Currently, it doesn't exist, this is a fact, and domain > is smb oriented, > isn't it? "domain" and "workgroup" are smb oriented. But other user management systems use a tree to locate users. I think a "user location" field is needed, you can name it like you want (domain, workgroup, OU, directory...). > So what next, fish:// (andships://). hmmm... > I mean, i > think we should > provide a G.E.N.E.R.I.C structure, so i think we should > collect more auth > informations to know what is really needed. > Also, why no a chained sub-structure like > typedef struct auth_prop_s auth_prop_t; > typedef struct auth_prop_s { > char *prop_name; > char *prop_value; > auth_prop_t *next; > }; you end with something more complex than the gnome stuff. > (which is problematic due of memory allocation, so an > array of auth_prop_t > could be better) > > This way we can provide a generic way to pass > properties. > > Cheers. > - -- > 73's de Daniel "Der Schreckliche", F1RMB. > > -=- Daniel Caujolle-Bert -=- > seg...@cl... -=- > -=- http://naboo.homelinux.org > -=- Thibaut |