johnt@quintus.UUCP (John Grant) (02/26/91)
i) Is there some suitable way of dealing with temporary files ? What I am looking for is some automated way of dealing with them so that should a program terminate in some unfortunate manner, you don't slowly consume all available disk space with dead temporary files. Users of Word, Fullwrite etc etc would appreciate this - the need for "helper" inits would then be avoided. Better would be a method whereby the temporary files would stay around during one reboot, so that the user might be able to recover any data from them before the files are deleted on the next reboot. It strikes me that this functionality would be easy and inexpensive to implement. ii) I have heard that an, albeit small, RAM cache cannot be disabled. I would like someone to explain why we are *forced* to use this in a synchronous I/O environment - this 32K (?) is probably already implemented in the drive controller as a track cache, so why duplicate this functionality - doesn't this also hurt the performance of those NUBUS SCSI controllers with all the RAM ?
lsr@Apple.COM (Larry Rosenstein) (02/27/91)
In article <1492@quintus.UUCP> johnt@quintus.UUCP (John Grant) writes: >i) Is there some suitable way of dealing with temporary files ? What > I am looking for is some automated way of dealing with them so that > should a program terminate in some unfortunate manner, you There is a special folder for such temporary items. An application can locate it using the Folder Manager. Any files contained in this folder when the machine boots will be put into the trash in a folder called "Rescued Items from <disk>". You can either retrieve the files (in case they can be recovered) or you can empty the trash. -- Larry Rosenstein, Object Specialist Apple Computer, Inc. 20525 Mariani Ave, MS 3-PK Cupertino, CA 95014 AppleLink:Rosenstein1 domain:lsr@Apple.COM UUCP:{sun,voder,nsc,decwrl}!apple!lsr
tecot@momenta.com (Ed Tecot) (02/28/91)
johnt@quintus.UUCP (John Grant) writes: >i) Is there some suitable way of dealing with temporary files ? What > I am looking for is some automated way of dealing with them so that > should a program terminate in some unfortunate manner, you > don't slowly consume all available disk space with dead temporary > files. Users of Word, Fullwrite etc etc would appreciate this - the > need for "helper" inits would then be avoided. > Better would be a method whereby the temporary files would stay > around during one reboot, so that the user might be able to > recover any data from them before the files are deleted on the > next reboot. It strikes me that this functionality would be easy > and inexpensive to implement. The Folder Manager has a temporary folder that is periodically sweeped of all closed files. >ii) I have heard that an, albeit small, RAM cache cannot be disabled. I > would like someone to explain why we are *forced* to use this in a > synchronous I/O environment - this 32K (?) is probably already > implemented in the drive controller as a track cache, so why > duplicate this functionality - doesn't this also hurt the > performance of those NUBUS SCSI controllers with all the RAM ? The minimum RAM cache is now 16K. Note that the RAM cache does things that track caches typically don't, such as caching volume and directory information which wouldn't be on the same track as your file. Also, the RAM cache replaces the ROM cache (RAM refers to where the cache code lives, not the cache), which used 12K in the system heap. Finally, the B*Tree Manager and the RAM cache share code; so supporting both methods of caching would result in significantly more work. _emt