home *** CD-ROM | disk | FTP | other *** search
- From: Julian Reschke <reschke@GOEDEL.UNI-MUENSTER.DE>
- Subject: Re: dreaddir/fxattr
- Date: Wed, 19 Jan 94 14:58:41 MET DST
- In-Reply-To: <9401191138.AA11388@issan.informatik.uni-dortmund.de>; from "Andreas Schwab" at Jan 19, 94 12:38 pm
-
- >
- > Julian Reschke <reschke@GOEDEL.UNI-MUENSTER.DE> writes:
- >
- > |> Below is a first draft:
- > |> Proposal for a combined call for Dreaddir/Fxattr:
- >
- > |> LONG Dreadattr( WORD len, LONG dirhandle, char *buf, WORD lflag, XATTR *xattr);
- >
- > |> len, dirhandle and buf are just like in Dreaddir. lflag and xattr
- > |> are the standard parameters for Fxattr.
- >
- > I don't think this is needed. Look at the latest version of GNU find.
- > It makes many optimizations regarding the need to stat() a file.
- > E.g., when running on a directory with no subdirectories, and you only
- > need the names of the files, it doesn't make any stat() calls. Such
-
- There are a lot of cases when you can't do that information. find -printf,
- du, desktops displaying file windows and so on.
-
- > optimizations will give you more than a new system call, which is also
- > difficult to bind into the existing dirent interface. (Of course, the
- > optimizations are only possible on MinixFS, where you have real link
- > counts for directories.)
-
- The new system call can be implemented as frontend for the other two
- calls. And it adds the possibility to optimize it in a future TOS.XFS.
-
- >
- > If you can fix it by software, why changing the kernel? :)
-
- That's the whole issue: it's ths OS interface which is needlessly slow
- and complicated.
-
- --
- ---------------------------------------------------
- Julian F. Reschke, Hensenstr. 142, D-48161 Muenster
- eMail: reschke@math.uni-muenster.de jr@ms.maus.de
- ___________________________________________________
-