(no subject)

[ewx]

I accidentally killed one of my userpics, so I thought I'd see if I could find it in my squid. No luck yet, but as part of the process I generated a page of all the userpics that my squid has seen recently, and LNR and I thought this was rather cool.

Beware; this page contains over 600 images, some of them animated. Do not blame me if your browser cannot cope! The URL is http://www.greenend.org.uk/rjk/junk/up.html.

Comments

[bopeepsheep.livejournal.com]

What's startling is how many of those I instantly recognise (put a username to). I think I spend too much time online!

[addedentry.livejournal.com]

Absolutely. I am reminded of the old TotL Human Virus Scanner (http://totl.net/VirusScanner/).

[ewx.livejournal.com]

*laugh* (http://totl.net/Urban/)

[bopeepsheep.livejournal.com]

It's really rather pretty, now I look at it longer. How did you generate it?

[ewx.livejournal.com]

zgrep userpic /var/log/squid/access.log*|awk '{print $7}'|sort -u|perl -ne 'chomp;print "<img src=\"$_\">\n"' and then Emacs to tidy it up into proper HTML.

[imc.livejournal.com]

This presumably generates links to images which have been seen rather than ones that have necessarily been cached. If one has been `killed' since it was last accessed it presumably generates a broken link (whereupon you scrabble around in the cache to find out what it was). Do I understand correctly? I don't know how LJ's numbering scheme works, but I assume it's not possible to modify a usericon in such a way that the old URL still points to it, as that would introduce inaccuracies into the result.

(I'll have to see if one can save the results of about:cache which is probably the only way of doing this sort of thing without a squid.)

[ewx.livejournal.com]

What happened was that I accidentally upload a new pic with the same keyword as an old one; now the old one is not there even if I change the keyword. I have no idea whether it re-uses the same URL for the same keyword, or made up a new URL.

Either my squid hasn't seen the old pic in the relevant period or it did re-use the URL since there are zero 404s for userpics in access.log after viewing that page.

[imc.livejournal.com]

I said...
I'll have to see if one can save the results of about:cache

Mozilla has no problem; Netscape 7 doesn't want to pop up the save dialogue but it will happily edit the page and save from there, though this takes several minutes of cpu time. I put the results here (http://users.comlab.ox.ac.uk/ian.collier/Misc/picshome.html) (home machine, 186 pics) and here (http://users.comlab.ox.ac.uk/ian.collier/Misc/picswork.html) (lab machine, 303 pics).

I have the same three images of yours cached in both places: penguin, grin and marvin.

BTW you could save a `chomp' and a `\n' by using perl -l.

[sparrowsion]

Note -- it's a bit silly to run this script for one's own squid after having looked at the original....

[ewx.livejournal.com]

Heh l-) You could combine the grep and the awk (zcatting the logs in if they're compressed) and make the awk terminate when it spotted the original URL. The squid I run at work doesn't keep access logs at all (except transiently when being debugged) so it'd be silly in an opposite way to use it there.

[mhk.livejournal.com]

What is your squid?

[ewx.livejournal.com]

It's web cache (http://www.squid-cache.org/Doc/FAQ/FAQ-1.html#ss1.1a).

[ewx.livejournal.com]

Ah, that should be "it's a web cache", rather.