Hey,
attached is a simple patch which enables using the 'wordchars' setting from the various filetype definition files also for auto-completion, i.e. to decide whether the entered character should start/continue auto-completion or not.
Previously, we always used the hard-coded default set of wordchars (except for LaTeX where we added a backslash in the code).
The patch retains the special LaTeX hack and also the default behaviour of using Geany's default set of wordchars if not specific in the filetype definition file of the current filetype.
Using special wordchars for auto-completion might be better for some languages. I was playing around with some Django template tags and there I noticed we ignore the wordchars in the filetype definition file for auto-completion. This patch might change the auto-completion behaviour if there are wordchars in a filetype definition file is set for a certain filetype. I would definitely add a note to the docs that the wordchars setting is also applied for auto-completion (just forgot it in the attached patch).
Any objections?
Regards, Enrico