PackageInstaller and handling script files

[sil2100]

There were no problems reported regarding this matter yet, but I presume some packages use this feature. Right now we treat them as normal files.

[kaliber]

I need this functionality. Is there any chance to implement it? This should not be difficult in my opinion.

[sil2100]

It shouldn't. I will take care of this in the nearest 3-4 days, since right now I have too much workload.

[sil2100]

I'm sorry for the delay, I only recently found enough time to start fixing this problem. Could anyone attach/paste a link to a real-life package using this functionality? I would like to test the changes.

[diver]

IIRC cpu_fix used this feature ​http://www.bebits.com/app/3999

[sil2100]

Patch fixing the invalid script file handling + some other modifications.

[sil2100]

This patch should do it. I used the occasion to export the installation procedure to another thread, since earlier the application was simply not responding during file copy. Checked it on the cpu_fix package.

Would be grateful if anyone could check if everything works fine with this patch. Thanks!

[korli]

A few things I noticed:

• ParseScript and RunScript should be prefixed with an underscore.
• if inflate_data fails, the allocated *script array is not deleted.
• code style problems, example PackageInstall.cpp line132: } <newline> else {
• the macro RETURN_MESSAGE() is a problem IMO. I would add a method which calls Install and post the message if needed.

[sil2100]

Thanks for reviewing the patch, korli!

I have fixed the bugs you have highlighted in the new patch attached below. I have added the deletion of *script memory in _ParseScript(), although it was not really necessary - the memory allocated in *scipt during _ParseScript() was later freed in the DoInstall() method (if, of course, the inflate_data function failed). But your approach seems more consistent in this case.

I am also sorry for the style problems - I am trying to proof read the code before submition, but it's really hard to get rid of some coding habits. Hope I did not miss anything.

[sil2100]

I noticed that the *script parameter was a reminder of some other, older idea that did not end up in the final patch - so I removed it and made script a local variable for the _ParseScript() method. Attaching modified patch (the same name as before).

[korli]

Another question, why adding "#include <Locker.h>" to PackageView.h ?

[sil2100]

Ah, that's another bogus leftover, sorry for that again. This was added because at first I wanted some parts of the new lock-needing functionality in PackageView, but changed my mind at the end. It seems I forgot to remove the include.

I'm re-uploading the patch once again.

[sil2100]

Fixed according to the notes provided by korli, thanks! Also fixed something I forgot about earlier. (v3)

[korli]

Applied patch in hrev36020.