Re: [platform-core-dev] possible File.setContent() latency issues on Lin

[John Arthorne <John_Arthorne@xxxxxxxxxx>]

To answer your question, no, we have
never seen this or had anything like this reported to us.  The only
problem we know of that is slightly related is that certain configurations
of Linux+Samba+ClearCase can report invalid timestamps on successive writes/reads
in a single threads.  The following bug report discusses in detail
and has a patch attached (the patch has not been applied because it is
not an appropriate solution in the general case):

https://bugs.eclipse.org/bugs/show_bug.cgi?id=16280
--


platform-core-dev-admin@xxxxxxxxxxx wrote on 01/09/2004
06:35:16 PM:

> Hi.
> I'm trying to resolve bug #46991 and just wanted to see if your team
has 
> ever encountered any issues with coherence in write/read operations
on a 
> file under the Linux JVM (1.4.1 IBM/Sun used in releng builds). I
wonder 
> if it's possible that the write to the file completes, yet an immediate
> read from the file (in the same thread, so it's synchronous) does
not 
> see the new contents.
> 
> More specifically, the test writes out a launch configuration (using
> File.setContents()) and then initiates a launch which reads the new
> configuration file contents. The test occasionaly fails with an 
> unexpected launch config variable (the working directory in this case).
> I have looked hard at other potential failure causes and have no better
> candidates than the write/read issue.
> 
> I have searched Sun's bug parade and did not see any reports related
to 
> this behavior.
> 
> Thanks for any hints or comments.
> 
> -Chris
> https://bugs.eclipse.org/bugs/show_bug.cgi?id=46991
> 
> _______________________________________________
> platform-core-dev mailing list
> platform-core-dev@xxxxxxxxxxx
> http://dev.eclipse.org/mailman/listinfo/platform-core-dev