Bug 168509 – [SOUP] Stop removing the fragment identifier from resource requests

• Status: RESOLVED
• Resolution: FIXED
• Component: Platform

[Carlos Garcia Campos]

I don't know why, but have always removed the fragment identifiers from the URL when creating the SoupURI that is passed to libsoup. Maybe it was a bug in a very old version of libsoup, but it doesn't look necessary anymore and it's causing several layout tests failures.

[Carlos Garcia Campos]

Created attachment 301918 [details]
Patch

[Carlos Garcia Campos]

Created attachment 301919 [details]
Patch

[Carlos Garcia Campos]

Committed r212548: <http://trac.webkit.org/changeset/212548>

[Martin Robinson]

Out of curiosity I tracked down this behavior to:
https://trac.webkit.org/changeset/41058

Here's the original comment:
    // GIO doesn't know how to handle refs and queries, so remove them 
    // TODO: use KURL.fileSystemPath after KURLGtk and FileSystemGtk are 
    // using GIO internally, and providing URIs instead of file paths

[Martin Robinson]

The origin of the behavior is here: https://trac.webkit.org/changeset/33594

[Carlos Garcia Campos]

(In reply to comment #4)
> Out of curiosity I tracked down this behavior to:
> https://trac.webkit.org/changeset/41058
> 
> Here's the original comment:
>     // GIO doesn't know how to handle refs and queries, so remove them 
>     // TODO: use KURL.fileSystemPath after KURLGtk and FileSystemGtk are 
>     // using GIO internally, and providing URIs instead of file paths

But I think this was only done for local files. This code was removed actually. For URLS we have always done it.

[Michael Catanzaro]

So this is why anchor links have always been buggy for us...? WTF. :(