ReplaceDotMess() switches all \ to / in URL. This is wrong, it should not touch them after # or ?.

After thinking more about this, I realized this could be a no-win situation. Clearly, ReplaceDotMess() should fix the problems in any part of the URL that is part of a file path/name. But it is impossible to tell for sure what is a path and what is not. Look at: http://some.com/foo?\cgi-bin\myquery;par1\bar;"\*" Quite obviously "\cgi-bin\myquery" is a path and conversion should happen. But what about "par1\bar"? It could be a path or not. As to '"\*"', it looks like it is just a string to be passed so we probably shouldn't touch it. Should read the URL spec again to see if it has anything to say...

I have a fix for this, waiting for review ... see bug 13311 ... it is part of a bigger patch to solve some more URL parsing problems. As to the additional comments, I think heikki is right, it is a no win situation. Using ReplaceDotMess only on the path (everything before ;?#) seems to me is doing the least damage.

fix checked in ...

thanks Andreas.

Bulk move of all Necko (to be deleted component) bugs to new Networking component.

Verified these in debugger, nsStdURL:GetSpec() with following URLs: target.html?/1/3#/5/6 target.html?/1/3#/5/6/slot('?#/') target.html?/1/3#/5/6/slot('\?#/') All the member variables seem to hold correct values.