Add API test for FileSystem::fileExists() on a broken symbolic link

• Component: Web Template Framework
• Status: RESOLVED FIXED

[Chris Dumez]

Add API test for FileSystem::fileExists() on a broken symbolic link.

[Chris Dumez]

Created attachment 427918 [details]
Patch

[Darin Adler]

Comment on attachment 427918 [details]
Patch

Should also add one for a non-broken symlink *to* a broken symlink

[Chris Dumez]

(In reply to Darin Adler from comment #2)
> Comment on attachment 427918 [details]
> Patch
> 
> Should also add one for a non-broken symlink *to* a broken symlink

For what it's worth FileSystem::exists() for a symlink to a *non-broken* symlink seems to return false. I have noticed this today. I will try and figure out why but the good? news is that it seems this is not new behavior from porting to std::filesystem.

[Darin Adler]

I suppose all these test cases show why we’d rather *not* have our own file system layer. But the convenience of using WTF::String directly is so tempting.

[Chris Dumez]

Comment on attachment 427918 [details]
Patch

Clearing flags on attachment: 427918

Committed r277113 (237411@main): <https://commits.webkit.org/237411@main>

[Chris Dumez]

All reviewed patches have been landed.  Closing bug.

[Radar WebKit Bug Importer]

<rdar://problem/77622567>

[Chris Dumez]

(In reply to Darin Adler from comment #4)
> I suppose all these test cases show why we’d rather *not* have our own file
> system layer. But the convenience of using WTF::String directly is so
> tempting.

I personally don't mind having our own FileSystem layer with WTF::String convenience and some other small niceties, as long as we keep the WTF::FileSystem layer cross-platform & small (by leveraging std::filesystem) and with a good set of tests.

The issue in my opinion was that we had 3 separate implementations with slightly different behaviors and almost no tests.

[Chris Dumez]

(In reply to Chris Dumez from comment #3)
> (In reply to Darin Adler from comment #2)
> > Comment on attachment 427918 [details]
> > Patch
> > 
> > Should also add one for a non-broken symlink *to* a broken symlink
> 
> For what it's worth FileSystem::exists() for a symlink to a *non-broken*
> symlink seems to return false. I have noticed this today. I will try and
> figure out why but the good? news is that it seems this is not new behavior
> from porting to std::filesystem.

access() on a symlink to a valid symlink returns -1 and errno is 2, which is:
[ENOENT]           The named file does not exist.

[Chris Dumez]

(In reply to Chris Dumez from comment #9)
> (In reply to Chris Dumez from comment #3)
> > (In reply to Darin Adler from comment #2)
> > > Comment on attachment 427918 [details]
> > > Patch
> > > 
> > > Should also add one for a non-broken symlink *to* a broken symlink
> > 
> > For what it's worth FileSystem::exists() for a symlink to a *non-broken*
> > symlink seems to return false. I have noticed this today. I will try and
> > figure out why but the good? news is that it seems this is not new behavior
> > from porting to std::filesystem.
> 
> access() on a symlink to a valid symlink returns -1 and errno is 2, which is:
> [ENOENT]           The named file does not exist.

Oh, it was an issue with the way FileSystemTest::SetUp() was constructing the symlink. I'll fix that and add more tests shortly.