On Monday 12 July 2004 15:26, Tuomo Valkonen wrote: [...] > However, Ion should respond to transient size change requests correctly (it > even stores the requested height) and this works at least with preferences > dialog of opera 5.0 that is really small initially. Maybe Eclipse is > calculating the height request incorrectly, being confused as the transient > doesn't have a frame? Does resizing the containing frame do anything? > Having just bought another 160GB of disk space for data, I still don't have > enough space in my apparently meager 5GB /usr to install Eclipse (what a > piece of bloat!) with all its dependencies.
Hey, it's not *that* big. This 3.0 installation is only 75MB. Um... ...okay. Moving on: > If you're willing to debug it, it might be a good idea to add some > printf:s to the clientwin_handle_configure_request function printing > out what kind of sizes the program is requesting. And if these turn > out to be ok, elsewhere in the file (clientwin_fit etc.) to see what > sizes are actually used and where. I'm not sure my lua skills are up to that. Also, I'm not entirely sure where ion's debugging information is going... I did add a kludge so that ion thinks that all Eclipse windows are toplevels, which mostly works. However, something is definitely screwy; certain types of window aren't sized correctly. What seems to happen is that you end up with the dialogue floating in the middle of the toplevel, the dialogue being the size of the *last* dialogue that got displayed... or worse; frequently I've ended up with 1x1 unresizeable windows and have only gotten out because I know what keys to press. Incidentally, this happens regardless of whether I'm on a floating or paned workspace. Is there a kludge to tell ion to ignore the program-specified maximum size? Too large is much better than too small, and if I can force the dialogues to be resizeable I should at least be able to get some use out of them. -- +- David Given --McQ-+ | [EMAIL PROTECTED] | C:\DOS, C:\DOS\RUN, RUN\DOS\RUN | ([EMAIL PROTECTED]) | +- www.cowlark.com --+
