dirstate2 comments

[John Arbash Meinel]

-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1


...

> 
> I wonder how hard, or how bad it would be for performance, to just
> throw away stat cache updates during a logical read lock.  It might be
> an interesting experiment, and it seems like it might not be much
> slower in realistic cases.  It would avoid an amount of complexity to
> do with these low-priority transactions, and it would avoid some
> operations that we know are risky on network filesystems that aren't
> quite coherent.
> 
> 

If we fix all the bugs where write operations currently are *not*
updating the stat cache, it would probably be ok.

John
=:->
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.4.9 (Cygwin)
Comment: Using GnuPG with Mozilla - http://enigmail.mozdev.org/

iEYEARECAAYFAkwDyxwACgkQJdeBCYSNAAO/zwCeOlYG49adsZugUToovQjfP6l+
fnQAoIvetIMnHs1x/4ky4f/79djNDNxe
=+nCs
-----END PGP SIGNATURE-----