This implementation doesn't create any more filelist objects than were
created before. Exactly one per pkgplan.
I agree with you about the use of the content cache - that wasn't in
place at the time I wrote the fix. However, looking up the file in the
content cache is more expensive than just adding the action to the list.
As I said before, the only extra memory cost is the list. The action
objects are in memory already anyway.
Tom
[EMAIL PROTECTED] wrote:
Tom,
filelist.py:
- line 111-113: You've taken out a check that we use to incrementally
download files as we fill up a filelist. The while is here in case
we don't successfully download all of the files in the first try.
Yes, this is intentional. In fact, I was surprised that filelist started
downloading files before all of the actions had been processed, as this
misses the opportunity to download files with the same hashval only once
(unless they happen to be close to each other in the manifest).
We keep the downloaded files in a content cache now. If we've
downloaded a file once during a transaction, we're not going to download
it again. Generating a huge number of filelist objects and then issuing
the download in a separate pass is both unnecessary and likely to bloat
our memory footprint.
-j
begin:vcard
fn:Tom Mueller
n:Mueller;Tom
org:Sun Microsystems, Inc.;Update Center Software
adr:;;21915 Hillandale Dr;Elkhorn;NE;68022;USA
email;internet:[EMAIL PROTECTED]
title:Senior Staff Engineer
tel;work:877-250-4011
tel;fax:877-250-4011
tel;home:402-916-9943
x-mozilla-html:TRUE
version:2.1
end:vcard
_______________________________________________
pkg-discuss mailing list
[email protected]
http://mail.opensolaris.org/mailman/listinfo/pkg-discuss
