On 12/20/2010 7:40 PM, Pavel Sanda wrote:
although all the advantages above look like very good step forward i'm afraid there is nobody around who is going to write new NSIS code (i suppose Uwe is not going to leave his code for some unfinished stuff he is not familiar with.)
That's why I think more people should be involved so that Uwe, myself and maybe some others will at least be reasonable familiar with the code.
so more direct question would be whether you intend at least continue the old standard installer...
I find the current situation with different types of installers way too confusing for users. I don't think it's helping the LyX project. There are features in the standard installer which I think are import such as not requiring administrator privileges. Also the way I bundle external applications (ImageMagick/Ghostscript) I think is more reliable. They are private to LyX, I don't modify registry keys that are supposed to belong to the original applications. But the disadvantage of having two different installers might outweigh this.
I still believe the best way forward is to start with the CMake build tools and the new libraries I've recently ported to Visual C++ (ftp://ftp.devel.lyx.org/pub/contrib/windows/). Then LyX can be compiled as a real Windows GUI application. Based on that it shouldn't be too hard to create an easy to understand installer script. It would require some major updates anyway to get rid of the old launcher process and to replace the spell checking / thesaurus libraries. So starting with a new script would be a sensible thing to do.
The question is whether this is feasible for 2.0. Perhaps I could start with build instructions and a really basic installer script that does only what's necessary to install LyX itself. From there we could continue to add new features. However before doing anything like that we have to be sure everyone agrees with this approach.
Joost
