Please ignore my last mail. I get this mail from you at the same
time I send mine.
>It looks like there's still a bit of a problem with the MRUMemoryStore.
>It's only happening if the fs store is used.
Seems like.
>The reading and writing of cached files are looking in different places.
>Basically, the calls to fsstore.get() and fsstore.store() are using
>different parameters. The call to fsstore.store() is adding in
>the cachedir path and the call to fsstore.get() isn't.
Yes that is right, it is careless bug from me.
>I've attached a Q&D patch to this and it works ...
Thanx, but you forgot to fix the remove() method.
>BUT I've also noticed that the cache keys are used by other methods
>such as remove().
yes you're right
>This patch is assuming that the key and the file on the filesystem are in
fact
>the same thing. Is that the intent? The keys now being passed to the
store()
>and get() methods look like this:
>cache-dir-prefix/PCK%3ACCK%3A3-html-1PCK%3...
>rather than just
>PCK%3ACCK%3A3-html-1PCK%3...
>Other methods would not be getting the cache-dir-prefix/ . Are the
>cache keys intended to look like filesystem-like things? Or are
>the keys supposed to be some abstraction of that?
no no, this really a bug. I fixed yesterday the path determination and
forgot
to append it on the other methods.
>Anyway, here's the quick & dirty patch that'll get the fs store working.
Thank you very much. But I would like to send a patch with the whole fixes
when this is ok.
---------------------------------------------------------------------
To unsubscribe, e-mail: [EMAIL PROTECTED]
For additional commands, email: [EMAIL PROTECTED]
