On Wed, Sep 06, 2017 at 10:14:34AM +0200, Florian Pritz via aur-general wrote: > On 05.09.2017 14:07, Morten Linderud wrote: > > signoff[2] is a tool I have written that helps testers with signing off on > > packages they have installed from testing. It comes with neat > > auto-completions > > and enough commands that it should replace the signoff page. Several > > testers are > > using this to signoff packages in the testing repositories. > > I've once wrote something similar[1]. I haven't used it lately, but it > still seems to work. I could have probably done a better job at > promoting it. > I believe i was made aware of your script after i started writing this, it's pretty neat
> Looking at your script, it's mostly the same, except you send a HEAD > request before fetching the signoff json. Since that json is probably > generated on demand, the HEAD request creates nearly equal load to a > normal one, minus actually sending and possibly compressing the json. > After all, the server needs to generate the page to know how long the > content is. > > It's possible that archweb caches the generated json and my hunch is > wrong, but it might be worth looking at that. > > What can really reduce load are the ETag and/or If-Modified-Since > headers, but I don't know if archweb supports those. > The caching issue has crossed my mind a few times. I disabled compression on the HEAD request as it was added after i wrote the machanism, and it broke the content-length check. I'll take a look at the ETag/If-Modified-Since headers, thanks! > Also the get_packages function has a "tries" parameter that doesn't > appear to be used in any way. > Hurm, unsure why that was added. Thanks for the heads up. -- Morten Linderud PGP: 9C02FF419FECBE16
signature.asc
Description: PGP signature