Re: [Nautilus-list] Small optimization

[Darin Adler <darin bentspoon com>]

On Friday, June 29, 2001, at 01:16  PM, Ettore Perazzoli wrote:

> I have started playing with Nautilus last night, hoping to maybe
> contribute something useful to optimize it a bit, and I came across
> get_metafile() in nautilus-directory-metafile.c, which seems to be doing
> a useless alloc/free and CORBA_exception_init/free in the case where you
> already have the CORBA interface set up.
>
>   This gets called for each file when opening a directory, and the CORBA
> interface is already up in that case, so I think it should be fixed.

I'm not sure I'd consider this a real optimization. I'd bet that it doesn'
t make a measurable difference; it's usually a mistake to "optimize" 
existing code in a way that users can't detect.

On the other hand, since it's annoying to have code that does extra stuff 
I don't see a compelling reason to leave it the way it is.

>   Can I commit this patch?

The concept of the patch is just fine, but it doesn't follow Nautilus 
coding style guidelines. As described in nautilus/doc/style-guide.html, we 
put all declarations at the top of the function, so the declarations 
should be left outside the if statement.

Otherwise the patch looks fine and committing would be fine either way; 
they are just guidelines and we can change them if we like, too.

>   BTW, what is the reason for having a hard limit on the number of files
> that can be displayed in a Nautilus directory view
> (NAUTILUS_DIRECTORY_FILE_LIST_HARD_LIMIT)?

I believe the reason had to do with major performance problems. Without 
the hard limit, if you did a search that returned most of the files on 
your disk (like searching for *.*) Nautilus would try to display a window 
that contains them all and come to a dead halt.

But I think a hard limit is not a great idea, and it would be nice if 
there was some other solution to this problem.

    -- Darin