On Sat, 10 Aug 2013 23:12:02 +0200 Roland Mainz wrote:
> On Sat, Aug 10, 2013 at 6:00 PM, Irek Szczesniak <[email protected]> 
> wrote:
> > On Sat, Aug 10, 2013 at 5:47 PM, Glenn Fowler <[email protected]> wrote:
> >>
> >> On Sat, 10 Aug 2013 17:43:25 +0200 Irek Szczesniak wrote:
> >>> On Fri, Aug 9, 2013 at 6:58 PM, Glenn Fowler <[email protected]> 
> >>> wrote:
> >>> > this will be in the next alpha as
> >>> >         /dev/fcntl@...
> >>> > because it works on all types of paths, even /dev/fd
> >>> >
> >>> > the patch happened quickly because the background work in pathdev() and
> >>> > syscall restart intercepts (both part of the stability/debugging flurry 
> >>> > this summer)
> >>> > were already in place
> >>> >
> >>> > it handles relative paths like this
> >>> >         /dev/fcntl@direct/./file_in_pwd
> >>
> >>> I don't like the name. It implies that all fcntl operations are
> >>> possible which is not the case. For example, how do you pass the
> >>> return value of fcntl() back to the user and what does it mean? How do
> >>> you implement locks, shares, leases with /dev/fcntl? The name is
> >>> misleading.
> >>
> >> fair enough
> >> how about "oflags"
> >
> > apropos oflag yields nothing. /dev/file was better than that one.
> > /dev/open would be a choice but that's already taken:
> > ls -l /dev/open
> > lrwxrwxrwx   1 root  root    46 Apr  5  2010 /dev/open ->
> > ../../devices/pci@0,0/pci1022,1102@18,2:mc-open

> Grumpf... that's why I proposed /dev/file@ ... all other similar stuff
> in /dev is already used-up by Solaris or Linux devices. Crazy enough
> there is a /dev/ioctl but no /dev/fcntl ... but I don't like
> /dev/fcntl very much because the *only* purpose is to open _files_
> with extra flags (remember the only reason we have this was to get
> access to |O_DIRECT| and |O_SYNC| for HPC (=high performance
> computing) purposes to circumvent/bypass limitations in other
> applications which don't have special flags to use this flags
> directly.
> |O_DIRECTORY| was later tacked-on because I tought it's usefull and
> prevents mishaps in the #ifdef logic of the _original_ prototype patch
> (before I added |ENOSYS|/|EINVAL| later) since it assumed that at
> least one |O_*| flag must be there.

> Can we stick with /dev/file@, please ?

fine

_______________________________________________
ast-developers mailing list
[email protected]
http://lists.research.att.com/mailman/listinfo/ast-developers

Reply via email to