Comment 10 for bug 832061

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

> ]> I don't think the locking stuff needs to be configurable. We ]>
> could, but I think you're far of in diminishing returns at that ]>
> point. (We've had the same value for ~3-5 years now.)
>
> Well, consider the issue we encountered a couple of times with
> mounted file systems on windows where we couldn't get enough data
> from the user(s) able to reproduce the issue. IIRC, you and Martin
> gives quite precise guidance about which code should be changed to
> acquire the data. Despite your efforts, we're still waiting for
> data.

Actually, not related at all. The failures on Windows are because
"os.rename(src, target); open(target) # failing". Nothing configurable
that could have been done, because it was an unexpected case.

If we add a "try again after X milliseconds", yes, that could be
configurable.

John
=:->

>
> My point here is that if the infrastructure is in place we just
> have to ask for a command to be executed, no code to write, no code
> to install.
>
> Is there something wrong with that ?
>
> I don't expect all such cases to be covered, but everywhere we
> decided to define a constant, we already considered that some other
> value(s) *may* potentially be useful (right ?), this bug aims at
> making them reachable by a simple command. Nothing more.
>
> Hence my previous question: where are the others ? If there is
> none left, fine, let's close the bug, we can even close it now and
> re-open it if needed.
>

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

iEYEARECAAYFAk5Y44AACgkQJdeBCYSNAAN69QCffPmtwCdQAlRCmYxwzoss2L4h
hN0AoKZ7/2H2j1Yum7eFZYfGE5KwbrIz
=uUkT
-----END PGP SIGNATURE-----