peter@sugar.UUCP (Peter da Silva) (07/17/87)
We just bought FACC, and immediately noticed a few things that are, um, not optimal. It's really cool, and I understand a new version (FACC II?) is coming out, but... Point. Facc does not allow the number of buffers to be specified ahead of time, a problem for CLI fans who don't want to have to lean on the "more" gadget whenever it comes up. How about "FACC +500K"? It takes a while to get the buffers up to 1700. Point. Facc doesn't background itself. We can't think of a good reason to run Facc from the CLI other than background, so why not do it anyway. After all, what's 48 bytes compared to 128K? Question. When Facc caches stuff, does it cache the whole track read in when the block is read? If not, is there any good reason why it shouldn't? If so, is there any way to defeat it for folks with little 512K machines? Question. Does FACC II come up with a window by default? We would prefer that it just be runnable from the startup-sequence without bragging. Question. Does FACC preferentially cache directory and file header blocks. If not, I'd like to suggest that FACC II do so. I remember reading something about someone's idea for making FACC a disk cache rather than a drive cache. That seems an admirable idea, and would make any sort of RAM disk far less important. Is it possible without total rewrite? It's depressing to watch 1600 buffers go out the window when you pop a disk. I like the dithered background to the FACC window. A nice touch. Making the window borders the standard color for such things would be an even nicer one. Off the top of my head I'd like to propose a new "law of computing", to wit da Silva's rule #3: don't do things differently just for the sake of being different... product differentiation isn't worth it. Or, why is Marauder so fancy when all I want to do is back up disks? -- -- Peter da Silva `-_-' ...!seismo!soma!uhnix1!sugar!peter (I said, NO PHOTOS!)
cjp@vax135.UUCP (Charles Poirier) (07/20/87)
In article <405@sugar.UUCP> peter@sugar.UUCP (Peter da Silva) writes: >Point. Facc does not allow the number of buffers to be specified ahead >Point. Facc doesn't background itself. We can't think of a good reason to >Question. Does FACC II come up with a window by default? We would prefer >Question. Does FACC preferentially cache directory and file header blocks. These things and more are handled in FACC II. Remember that ASDG promises free lifetime upgrades on FACC, so there's no point in waiting for FACC II's release. >Question. When Facc caches stuff, does it cache the whole track read in > when the block is read?.... I don't know this. >I remember reading something about someone's idea for making FACC a disk >cache rather than a drive cache. That seems an admirable idea, and would >make any sort of RAM disk far less important. Is it possible without total >rewrite? It's depressing to watch 1600 buffers go out the window when you >pop a disk. My idea, and I've bashed on Perry about it. But it would, in fact, require a total rewrite. FACC installs itself at a point where the volume name is not in the I/O request, only a drive unit and block number. -- Charles Poirier (decvax,ucbvax,ihnp4,attmail)!vax135!cjp "Docking complete... Docking complete... Docking complete..."
perry@atux01.UUCP (Perry S. Kivolowitz) (07/21/87)
In article <405@sugar.UUCP>, peter@sugar.UUCP (Peter da Silva) writes: > We just bought FACC, and immediately noticed a few things that are, um, > not optimal. It's really cool, and I understand a new version (FACC II?) > is coming out, but... Peter, many of the things you mentioned were in fact mentioned before (but thank you none-the-less) and have been accounted for in FaccII. I will be formally premiering FaccII at the BCS meeting in Boston Tuesday night (but Jaug did get a sneak preview). Concerning making Facc a disk cache rather than a drive cache: It can be done but it is really not easy. The driver (the level at which Facc does its work) is asked to read sector number of unit N not sector number of disk ``Workbench''. There is no straight forward means for getting around that but as I said, I believe I might have a workable means. If so, then you'll see it become part of the Facc evolutionary path. The FaccII programmer's manual is available from us at no charge if you send a stamped self addressed 8 x 10 inch envelope or larger. Using the FaccII programmer's interface, you can write applications which can control the many new (and of course the old) features re- motely. If the net is interested, I can post a complete run down of new FaccII capabilities and features. Perry S. Kivolowitz - ASDG Incorporated - (201) 563-0529