Hi Jayant, > We had earlier created "dlopen hack" for evo-lib dependancy. Can we > remove it now!!
For the evo-lib, I don't consider usage of dlopen a hack. This is because for a given system, it's more likely for glib to be present than the evo lib. If we link the driver against the evo lib, then there is no diagnostics possibility on a system without evo, since the driver will simply not load at runtime. If we don't link against the evo lib, then the driver can at least be loaded, and has more possiblities to give the user a reasonable feedback when she attempts a connection. To the user, this might be the difference between "Evolution does not appear in the UI at all, so I don't even know about it", and "when I select Evolution, it gives me a message like 'No suitable Evolution version was found'". This may not sounds too much, but personally I'd consider it worth it :) (the more since the code is already there now.) Not to mention that before glibc 2.2.5, there's a bug in the library loader which causes subsequent calls to dlopen to *crash*, if you had one call which failed. So the more likely any library fails to load, the more likely OOo will crash a minute later. Ciao Frank -- - Frank SchÃnheit, Software Engineer [EMAIL PROTECTED] - - Sun Microsystems, Inc. http://www.sun.com - - OpenOffice.org Database Access http://dba.openoffice.org - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - _______________________________________________ evolution-hackers maillist - [email protected] http://lists.ximian.com/mailman/listinfo/evolution-hackers
