[20686] in Athena Bugs
Re: mozilla cache location
daemon@ATHENA.MIT.EDU (t. belton)
Thu Sep 5 13:20:17 2002
Date: Thu, 5 Sep 2002 13:20:15 -0400 (EDT)
From: "t. belton" <tbelton@MIT.EDU>
To: Jonathon Weiss <jweiss@MIT.EDU>
cc: <bug-infoagents@MIT.EDU>
In-Reply-To: <200209050140.VAA26070@the-other-woman.mit.edu>
Message-ID: <Pine.GSO.4.33L.0209051308360.19204-100000@iphigenia.mit.edu>
MIME-Version: 1.0
Content-Type: TEXT/PLAIN; charset=US-ASCII
This is deliberate. I went to some trouble to make sure that when Mozilla
converts your 4.x prefs, the cache pref is NOT converted. The reason for
this is that Mozilla handles cache location differently from netscape 4,
and what was useful there is not useful here.
The actual files will be stored in /var/tmp/Mozilla/Cache. Mozilla must
have control over the inner Cache directory, including being allowed to
create it itself. OTOH it will NOT create /var/tmp/Mozilla or any other
parent location, and this must be handled by the wrapper or Mozilla will
choke and die when it tries to create Cache.
As for the cache location not being user independent: On public
workstations the cache is erased by the wrapper after the user exits
Mozilla anyway. That takes care of the privacy issues.
Given Mozilla's own protections against multiple instances running on the
same machine (we will be using -remote to get around some of this, but
always for a single profile/user, not multiple users), I can't think of
any sharing issues - I can't think of a case where two different
users/profiles would be writing into this cache at the same time. Mozilla
itself, as I say, goes to some lengths to prevent this.
Given that I could name the cache parent directory anything I want, why
didn't I just stick to the existing scheme? Because writing the username
into the cache directory name would have been tricky to implement on first
run of Mozilla - due to the usual complication about not being able to
finess a preferences file that doesn't exist yet. I didn't want this cache
in the wrong place, not for the first run, not for any run. This way I can
put it in the globals and it works all the time, every time.
As for the disk cache size, that is Mozilla's internal default. In fact I
am not clear if this setting works properly; I have seen some machines
where this preference appears to be set to zero in prefs.js, but still
comes up as 50000. Setting it to a smaller non-zero value works. If you
think a smaller non-zero size is optimal I can put that in the globals and
see if it works.
On Wed, 4 Sep 2002, Jonathon Weiss wrote:
>
> Hi,
>
> I'm not sure where you wanted bug reports for Mozilla in infoagents,
> but this seemesd like as good a guess as any.
>
> My cache preferences were not properly converted from netscape, when I
> first started Mozilla.
>
> Under Netscape I had a 4000K memory cache, and a 20000K disk cache in
> /var/tmp/.netscape-cache-jweiss. Under Mozilla I ended up with a
> 4096K memory cache (sure, whatever) and a 50000K disk cache in
> /var/tmp/Mozilla. I don;t really think the extra size will be a
> problem, but the location is not user independant which is a problem.
>
> Jonathon
>
