Page 2 of 2
Posted: Fri Mar 05, 2010 4:09 pm
I tested the plug-in under Windows 7 but didn't noticed any problems so far. Do you have a (small) setup which can be used to reproduce the failure? If so, please send me this setup (including the source) to info[at]casic.de (10MB limit).
Thank you very much in advance for your support,
Posted: Mon Mar 08, 2010 4:47 am
My particular problem was caused by a version conflict of the installed font. I installed arialuni.ttf (v1.00), but Office 2007/2010 is using this font also, while their version is v1.01. For some reason, the Win7 font system was failing because fntcache.dat was not rebuild for the new (old) version of that file.
This was fixable by deleting fntcache.dat and a reboot, of including v1.01 in my installer.
It's curious that WinXP/Vista in the same situation didn't suffer this problem.
Posted: Mon Mar 08, 2010 10:29 am
Hi Markus and Gunter,
Thanks for the quick responses. Unfortunately it doesn't look like the problems we're seeing are quite the same as Gunter's, however I did stumble on to a few things that might help solve the problem. I tried putting together an extremely stripped-down installer this morning to send to Markus, but strangely enough that one worked perfectly, even on the same VM where the other installer fails. In the course of testing, I noticed that one difference between our installer and that used by the test project is that we sign our installer with a manifest so that it must be run as an administrator. I tried running the test installer as administrator, but again it seemed to work normally. The other major difference is that our installer does some slightly unusual things to put itself into 64-bit mode, and therefore installs to the C:\\Program Files folder, where the test app installed to C:\\Program Files (x86). I did confirm that the font file exists in the location we're attempting to install it from by adding a messagebox listing the path immediately preceding the install fonts line in the msicode. The messagebox displays as expected, leading me to believe we are indeed hitting the plugin with the correct path, although I have no way to confirm this.