[
Date Prev][
Date Next][
Thread Prev][
Thread Next][
Date Index][
Thread Index]
[
List Home]
|
AW: [geclipse-dev] AW: [Bug 228185] BufferOverflowException duringopening output file for files from gLite sandbox
|
> But because Eclipse call fetchInfo() also from
> decorators/label-providers I have no chance to avoid asking for
> auth-token during starting g-Eclipse.
Yes, that is exactly the problem, stupid, isn't it?! Nevertheless since
all these calls are filtered by the gecl stuff we at least have the
chance to catch them before they enter the native EFS implementations
like gsiftp (like it is already done with the childNames() method). But
as stated before this means mature reworks of the gecl stuff and is no
option for this release.
> I think I will pass flag FORCE to fetchInfo(), and when this flag is
set
> I will force to download current IFileInfo from the server, otherwise
I
> will return cached IFileInfo.
Which FORCE? There is not FORCE defined in EFS?!
> Any better ideas are highly welcomed!
Well, not for a quick solution. Nevertheless we still have the option to
remove the new opening strategy and to replace it with the old one until
the next release. Replacing this with the old one is a one-liner, so no
problems here. In the meantime I would prefer this solution. We could
then make everything ready for M6 again.
So at least the WP leaders are now asked to vote ASAP:
Should we
1) replace the new opening strategy with the old one and leave the
gsiftp and srm implementations as they are for M5?
Or should we
2) leave the new opening strategy as it is and implement Mariusz's quick
and dirty solution with the FORCE parameter?
So from me this is a +1 for 1) and a 0 for 2). What about the others?
The decision should be made ASAP in order to allow the nightly build to
reflect the corresponding changes.
Cheers, Mathias
