Jean-Marc Lasgouttes wrote: > Angus> Interesting. I found that I had to start lyx from the > Angus> /home/Angus/qt3/lib directory (so that qt333.dll was found; I > Angus> could have changed my PATH variable but only found that out > Angus> later) > > Well, in a normal operation, qt333.dll would be in the lyx directory, > anyway.
Why? I think I'll end up installing Qt in "C:\Program Files\Qt" and put the headers in "C:\Program Files\Qt\include". Things other than LyX can then use it... > Angus> lyx -sysdir /home/Angus/lyx-13x/lib \ -userdir > Angus> /home/Angus/lyx-13x/build-qt/lib > > Angus> so that chkconfig.ltx and textclass.lst could be found. I guess > Angus> that this reinforces your points above. > > There is special code for LyX/Mac that detect running from an > application bundle. That's not very nice, but it works. > > I am surprised that you had to specify -userdir. Actually, it seems > like a very strange thing to do. I found that LyX would refuse to start if it couldn't find chkconfig.ltx (-sysdir) or textclass.lst (-userdir). However, I didn't investigate at all. Like I said the other day, I was just very happy to get LyX appearing on the screen. I didn't actually try and use it... My plan with Ruurd's stuff is to commit a few more bits and bobs from the linux machine and then to fire up the Windows machine, merge the remainder of the patch back into the local tree and try and get the thing to compile again. The primary goal is to get the sources to compile under Windows without needing to #include "os_win32.h" from os.h. (Individual .C files will still need os_win32.h.) (This strategy is driven by the fact that the MinGW gcc is INCREDIBLY slow.) When we get to that stage, I'll post the remaining bits of Ruurd's changes so we can gather our thoughts about the best way to proceed to a fully-functional Win32 LyX. -- Angus
