dorner@pequod.cso.uiuc.edu (Steve Dorner) (03/27/91)
>How about this: no more than 20 files open per process under UNIX >(hard-coded limit, not configurable)? Versus the configurable systemwide >number on the Mac (default 40). Depends on what you mean by configurable. It's a trivial thing to change if you have kernel source (not all that uncommon). I also think it's important that the limit is per process; much, much less restrictive than the mac's 'all processes' limit. As an application writer, I can know that, if I stay within my 20 fd limit, everything is peachy. On the mac, I *ought* to worry about leaving some of those precious fd's for other processes (some things, notably the MPW linker, don't bother with such niceties). -- Steve Dorner, U of Illinois Computing Services Office Internet: s-dorner@uiuc.edu UUCP: uunet!uiucuxc!uiuc.edu!s-dorner
hamilton@kickapoo.cs.iastate.edu (Jon Hamilton) (03/28/91)
ldo@waikato.ac.nz (Lawrence D'Oliveiro, Waikato University) writes: >In article <1991Mar26.153602.276@ux1.cso.uiuc.edu>, dorner@pequod.cso.uiuc.edu >(Steve Dorner) challenges: >"...name one way you have to watch resource usage under UNIX that wouldn't >ALSO apply to the Macintosh." >How about this: no more than 20 files open per process under UNIX >(hard-coded limit, not configurable)? Versus the configurable systemwide >number on the Mac (default 40). That's odd - I can configure it under A/UX... >Has this limit been lifted or loosened up in current versions of UNIX? >Lawrence D'Oliveiro fone: +64-71-562-889 >Computer Services Dept fax: +64-71-384-066 >University of Waikato electric mail: ldo@waikato.ac.nz >Hamilton, New Zealand 37^ 47' 26" S, 175^ 19' 7" E, GMT+12:00 >Those who make remarks about expanded and extended memory under DOS >should remember that the idea of using a device driver to access >memory originated with UNIX. -- Jon Hamilton hamilton@kickapoo.cs.iastate.edu " I feel a lot more like I do now that I did before I got here " - can't remember who
ldo@waikato.ac.nz (Lawrence D'Oliveiro, Waikato University) (03/28/91)
In article <1991Mar26.153602.276@ux1.cso.uiuc.edu>, dorner@pequod.cso.uiuc.edu
(Steve Dorner) challenges:
"...name one way you have to watch resource usage under UNIX that wouldn't
ALSO apply to the Macintosh."
How about this: no more than 20 files open per process under UNIX
(hard-coded limit, not configurable)? Versus the configurable systemwide
number on the Mac (default 40).
Has this limit been lifted or loosened up in current versions of UNIX?
Lawrence D'Oliveiro fone: +64-71-562-889
Computer Services Dept fax: +64-71-384-066
University of Waikato electric mail: ldo@waikato.ac.nz
Hamilton, New Zealand 37^ 47' 26" S, 175^ 19' 7" E, GMT+12:00
Those who make remarks about expanded and extended memory under DOS
should remember that the idea of using a device driver to access
memory originated with UNIX.
kent@sunfs3.Camex.COM (Kent Borg) (03/30/91)
In article <1991Mar27.144323.21504@ux1.cso.uiuc.edu> dorner@pequod.cso.uiuc.edu (Steve Dorner) writes: >I also think it's important that the limit is per process; much, much less >restrictive than the mac's 'all processes' limit. As an application >writer, I can know that, if I stay within my 20 fd limit, everything >is peachy. On the mac, I *ought* to worry about leaving some of those >precious fd's for other processes (some things, notably the MPW linker, >don't bother with such niceties). I think that under 7.0 file control blocks are dynamically allocated by the system. I m sure there is still an upper limit, but it is likely memory capacity/heap fragmentation bound. If this is true, I like it better than either Unix's or 6.0's approach. (Do you really want to edit the source and rebuild your kernel to give a program more file control blocks? Somehow I don't think that Unix is going to be an important shrink-wrap market anytime soon.) -- Kent Borg internet: kent@camex.com AOL: kent borg H:(617) 776-6899 W:(617) 426-3577 "We foolishly did not realize that he was stupid." - April Glasbie 3-20-91
lsr@Apple.COM (Larry Rosenstein) (03/30/91)
In article <1897@camex.COM> kent@sunfs3.Camex.COM (Kent Borg) writes: > >I think that under 7.0 file control blocks are dynamically allocated >by the system. I m sure there is still an upper limit, but it is >likely memory capacity/heap fragmentation bound. I think is true also. There is a limit, which is imposed by the fact that refnums are signed offsets into the FCB table. This give a maximum of about 340 open files (since each FCB is 95 bytes). -- Larry Rosenstein, Object Specialist Apple Computer, Inc. 20525 Mariani Ave, MS 77-A Cupertino, CA 95014 AppleLink:Rosenstein1 domain:lsr@Apple.COM UUCP:{sun,voder,nsc,decwrl}!apple!lsr
ldo@waikato.ac.nz (Lawrence D'Oliveiro, Waikato University) (04/01/91)
In article <1897@camex.COM>, kent@sunfs3.Camex.COM (Kent Borg) mentions
that System 7.0 has a dynamically-allocated array of file control blocks
(FCBs), so that the number of simultaneously-open files is limited only by the
fact that FCB offsets must fit into a 16-bit word.
He says: "If this is true, I like it better than either Unix's or 6.0's
approach."
Not only is it true, Kent, but you can get the *same* functionality
under System 6.0.x today! All you need is a Control Panel module called
"Up Your FCBs", written by Pete Helme at Apple Computer. It's probably
available on the net somewhere.
Lawrence D'Oliveiro fone: +64-71-562-889
Computer Services Dept fax: +64-71-384-066
University of Waikato electric mail: ldo@waikato.ac.nz
Hamilton, New Zealand 37^ 47' 26" S, 175^ 19' 7" E, GMT+12:00
kent@sunfs3.Camex.COM (Kent Borg) (04/03/91)
In article <1991Apr1.222658.3289@waikato.ac.nz> ldo@waikato.ac.nz (Lawrence D'Oliveiro, Waikato University) writes: >In article <1897@camex.COM>, kent@sunfs3.Camex.COM (Kent Borg) mentions >that System 7.0 has a dynamically-allocated array of file control blocks >(FCBs), so that the number of simultaneously-open files is limited only by the >fact that FCB offsets must fit into a 16-bit word. 'Fraid not. Quote screwup time. I would *like* to have said that, but that part was actually written one of those smart Apple people. My part was just a vague and tentative comment that I thought FCBs were dymanic under 7.0. (Nevertheless, I do feel a bit smug to have been told that I was right and be backed up with impressive sounding details. Always like it when that happens.) -- Kent Borg internet: kent@camex.com AOL: kent borg H:(617) 776-6899 W:(617) 426-3577 "We foolishly did not realize that he was stupid." - April Glasbie 3-20-91