Re: [Gnu-arch-users] Some issues

[Matthieu Moy]

Andrew Suffield <address@hidden> writes:

> On Wed, Jun 09, 2004 at 11:36:34PM +0200, Matthieu Moy wrote:
>> Andrew Suffield <address@hidden> writes:
>>
>> > All of which is completely irrelevant. tla almost never reads all the
>>
>> tla almost never reads all the ids in a directory, except for
>>
>> tla commit
>> tla changes
>
> Wrong on both of those.

Then I have a buggy implementation of tla ;-) (or a buggy strace)

$ strace tla changes | & grep '\.arch-ids/.*\.id' | grep open
open("./.arch-ids/INSTALL.id", O_RDONLY) = 4
open("./.arch-ids/Makefile.in.id", O_RDONLY) = 4
open("./.arch-ids/configure.ac.id", O_RDONLY) = 4
open("./.arch-ids/install-sh.id", O_RDONLY) = 4
open("./.arch-ids/INSTALL.id", O_RDONLY) = 4
open("./.arch-ids/Makefile.in.id", O_RDONLY) = 4
open("./.arch-ids/configure.ac.id", O_RDONLY) = 4
open("./docs/.arch-ids/BINDINGS.id", O_RDONLY) = 4
open("./docs/.arch-ids/FEATURES.id", O_RDONLY) = 4
open("./docs/.arch-ids/HACKING.id", O_RDONLY) = 4
open("./docs/.arch-ids/TODO.id", O_RDONLY) = 4
open("./docs/.arch-ids/BINDINGS.id", O_RDONLY) = 4
open("./docs/.arch-ids/FEATURES.id", O_RDONLY) = 4
open("./docs/.arch-ids/HACKING.id", O_RDONLY) = 4
open("./docs/.arch-ids/TODO.id", O_RDONLY) = 4
open("./.arch-ids/install-sh.id", O_RDONLY) = 4
open("./lisp/.arch-ids/Makefile.in.id", O_RDONLY) = 4
open("./lisp/.arch-ids/xtla-browse.el.id", O_RDONLY) = 4
open("./lisp/.arch-ids/Makefile.in.id", O_RDONLY) = 4
open("./lisp/contrib/.arch-ids/ewoc.el.id", O_RDONLY) = 4
open("./lisp/contrib/.arch-ids/ewoc.el.id", O_RDONLY) = 4
open("./lisp/.arch-ids/xtla-browse.el.id", O_RDONLY) = 4
open("./lisp/.arch-ids/xtla-core.el.id", O_RDONLY) = -1 ENOENT (No such file or 
directory)
open("./lisp/.arch-ids/xtla-defs.el.id", O_RDONLY) = -1 ENOENT (No such file or 
directory)
open("./lisp/.arch-ids/xtla-xemacs.el.id", O_RDONLY) = -1 ENOENT (No such file 
or directory)
open("./lisp/.arch-ids/xtla.el.id", O_RDONLY) = -1 ENOENT (No such file or 
directory)
[...] 

I didn't benchmark tla to know how long opening id files was taking in
comparison to  the rest of  the operation, but  it _does_ scan  all id
files for a "tla changes".

>> tla tree-lint
>
> Misleading.
>
>> tla inventory
>
> And not usually run by hand.

And how does the fact that it isn't ran by hand impact performance?

--
Matthieu