Summary: | REGRESSION: navigator.plugins empty | ||||||||
---|---|---|---|---|---|---|---|---|---|
Product: | WebKit | Reporter: | mitz | ||||||
Component: | JavaScriptCore | Assignee: | Geoffrey Garen <ggaren> | ||||||
Status: | VERIFIED FIXED | ||||||||
Severity: | Major | CC: | eric | ||||||
Priority: | P1 | ||||||||
Version: | 420+ | ||||||||
Hardware: | Mac | ||||||||
OS: | OS X 10.4 | ||||||||
URL: | file:///Applications/Safari.app/Contents/Resources/English.lproj/Plug-ins.html | ||||||||
Attachments: |
|
Description
mitz
2005-12-16 04:59:20 PST
The regression is due to this change in kjs_navigator.cpp r1.41: - KConfig c(KGlobal::dirs()->saveLocation("data","nsplugins")+"/pluginsinfo"); + KConfig c(""); I should add that the simplest fix is to pass "pluginsinfo" to the constructor. Created attachment 5115 [details]
Fix
Mitz is right. Eric thought the old code passed a single string to a
constructor that returned NULL, so he replaced that string with "". In fact,
the old code passed one string to a constructor that returned NULL, and then
concatenated "pluginsinfo" to that string. So "pluginsinfo" is what we want.
I'm going to add "Plug-ins.html" to our layout test suite when I land this. Created attachment 5116 [details]
Fix
Previous patch had extra junk in it.
Comment on attachment 5116 [details]
Fix
Agreed. Mitz and I were discussting making this an enum, etc. but that can be
done later (in my next cleanup patch!). Thanks for fixing this. r=me
Oh, as a side-note. Plugins.html (your test) should check that navigator.plugins has above a certain number of plugins, or perhaps one or two which we ship by default. Unfortunately it can't just dump all the plugins as those will be different between machines. Thanks again. Thanks, Geoffrey! |