The scales files is opened NEW, then overwritten by REWINDing for each refinement cycle.
It is not explicitly closed, but relies on the compiler correctly flushing & closing the file on exit. This may be a compiler bug I suppose. I can change Scala to close the file explicitly, which might fix the problem Phil > Hi Phil, > > Looking at the file it is clear that the end is missing - the files are > identical to a certain point then the last few 0 values are missing and > there is no newline at the end of the ppc version... > > I have just run it again on the firewire disk and got the same error, > which is odd. So the problem looks like it is to do with the ppc mac not > the file system. Any ideas anyone? It would seem like a major operating > system fault to be missing (i.e. not writing) the end of a file... > > Thanks, > > Graeme > > -----Original Message----- > From: CCP4 bulletin board [mailto:[EMAIL PROTECTED] On Behalf Of > [EMAIL PROTECTED] > Sent: 20 March 2007 10:04 > To: CCP4BB@JISCMAIL.AC.UK > Subject: Re: [ccp4bb] OS X Scala SCALES file weirdness > > The SCALES file is plain text so you should be able to check it with an > editor > Phil > >> Hi Folks, >> >> Sorry this is slightly off topic. I have an OS X PPC mac mini and I > have >> noticed some very strange behaviour. When I run scala for one example >> the scales file that is produced is exactly 16384 bytes, and when I > try >> and restore this scala dies. Now running the same scaling job on an >> intel iMac the scales file is a few bytes larger and can be read >> appropriately. >> >> This suggests to me that a few bytes are being "lost". >> >> Weirder still, if I run the *same* job on an external firewire disk on >> the ppc mac it works fine and produces a slightly bigger than 16k > scales >> file. This suggests to me that there is something unhappy about the > file >> system on this device. This came with OS X 10.3 and an "upgrade" to > 10.4 >> instead of a 10.4 fresh install, though I have no idea why this would > be >> a problem. On both machines I am running the job in my home > directory... >> >> Has anyone noticed this behaviour? More usefully - has anyone got any >> idea of how to prevent it? >> >> Thanks, >> >> Graeme >> > >
