On Wed, Jun 22, 2016 at 05:19:19PM +0100, Simon McVittie wrote: > If there is a compelling advantage to splitting up libraries, of course, > by all means do so.
I have an example: gspell: https://wiki.gnome.org/Projects/gspell The non-GUI parts could be implemented in GIO, with an extension point to not hard-depend on Enchant. And the GUI parts could be implemented in GTK+. I created gspell as a separate library for a few reasons: With a GIO extension point, a new library would anyway need to be created. And I thought that dlopening a library was as bad for startup time as a hard dependency. Now I know that a GIO extension point would actually be slightly better. Be able to iterate on the API more freely, and have something working (and used in several apps) in less time. Now gspell-1 has a stable API, but nothing prevents me from bumping to gspell-2 to break the API, making the new major version parallel-installable. In GIO, it is not possible. Be the maintainer. If I wanted to integrate spell checking in GIO and GTK+, I would still be busy with that project. With a separate library, it was done in 6 months, working on and off on it (and part-time). Having the code self-contained. Having all the spell-checking related code in one repository. The biggest *.c file contains 1400 lines. Mixing all that code in GtkTextView would make the code less clear, in my opinion. -- Sébastien _______________________________________________ gtk-devel-list mailing list gtk-devel-list@gnome.org https://mail.gnome.org/mailman/listinfo/gtk-devel-list
