Re: Application startup performance

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]


Behdad Esfahbod <behdad.esfahbod@xxxxxxxxx> writes:

> I think what Akira is saying is that it's not enough to check the mtime of the
> toplevel directory.

No argument there; it should be checking the mtime of every directory in
the hierarchy.

> My understanding of the thread is that, to fix it, Akira added a readdir() at
> the end; whereas what you are saying (which I agree with), is that instead of
> the readdir(), we should just traverse the subdirectories using the subdir
> list in the cache files, and check the mtime of all the subdirs.

Right, that should be sufficient, and I thought that was how the code worked.

> All these said, since, I don't think there exists an atomic "check mtime &
> rename" syscall available, *any* solution will be racy.  It's just about being
> easy to hit or not.

Completely agreed; it's all about making the window small and making the
consequences minor. In this case, the consequence should only be that
the cache file is out of date, and that a future application will
discover that at startup and also try to fix it. Eventually, some
application should manage to construct an accurate cache once font
installation is done...


Attachment: signature.asc
Description: PGP signature

Fontconfig mailing list

[Index of Archives]     [Fedora Fonts]     [Fedora Users]     [Fedora Cloud]     [Kernel]     [Fedora Packaging]     [Fedora Desktop]     [PAM]     [Gimp Graphics Editor]     [Yosemite News]

  Powered by Linux