[Webkit-unassigned] [Bug 106140] Major performance improvement in Qt's PluginDatabase implementation

[bugzilla-daemon at webkit.org]

https://bugs.webkit.org/show_bug.cgi?id=106140





--- Comment #1 from David Faure <faure at kde.org>  2013-01-05 03:40:42 PST ---
Created an attachment (id=181434)
 --> (https://bugs.webkit.org/attachment.cgi?id=181434&action=review)
the patch

Now with the patch, oops.

I'm wondering if my approach is valid: googling for NP_Initialize and NP_GetValue API docs seems to indicate that the browser is supposed to call NP_Initialize before NP_GetValue? It appears to work fine, though, all plugins I have here return correct stuff (description and mimetypes) from NP_GetValue even without calling NP_Initialize.


Note that I had to remove "m_infoIsFromCache = false" so that ensurePluginLoaded() doesn't return m_isLoaded (which is false at that point, when called from PluginDatabase::pluginForMIMEType). Basically this is what PluginPackageWin.cpp does as well (not calling load from fetchInfo, and not setting m_infoIsFromCache to false), while PluginPackageMac.cpp calls load from fetchInfo and sets that bool to false (like the Qt code was doing before this patch). So I presume this is correct (it does fix the actual loading of plugins), even though I don't fully understand the m_infoIsFromCache logic.

-- 
Configure bugmail: https://bugs.webkit.org/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are the assignee for the bug.