On Fri, 26 May 2006, Russell E Owen wrote: > I suggest that the DC1 queue code could be simplified and made more > robust by insisting that files be ordered alphabetically, rather than > measuring the creation date and picking the earliest file.
We did talk about this approach when we first started working on this, but we went with a sort-on-creation date as the most straight-forward approach, given that time order that the input files arrive is what we are after. Using a naming convention just moves the problem out of the filesystem to our middleware. I don't expect the filesystem to confuse creation date times in the context of DC1; however, it could be an issue under a production, long-running context. However, we expect to use a much different queuing model in the future when we no longer rely on ram disks to pass data between stages. That is, I expect we will throw this code away. hope this helps, Ray _______________________________________________ LSST-data mailing list [email protected] http://www.lsstmail.org/mailman/listinfo/lsst-data
