I added some information on this new behavior here - https://github.com/apache/libcloud/blob/f122600d2adf181a9b100cdd552cd02979c5b1b9/docs/compute/pricing.rst#downloading-latest-pricing-data-from-an-s3-bucket
Keep in mind that those 3 files are not public yet. I plan to make them public and read-only in the near future once the rate limits are sorted out. On Sat, Jul 11, 2020 at 4:07 PM Tomaz Muraus <to...@apache.org> wrote: > Yeah, I would actually prefer a git repository so everything is version > controlled, etc., but I went with the fastest and simplest approach > possible. > > I'm not exactly sure what the ASF rules are for something like that (I > would need to ask ASF infra team to create a new repo, create a bot account > which we could use in our CI, etc.) and that would likely take much longer > than the approach I went with. > > As far as libraries such as pytz (and to some extent also certifi) go - I > would say it's a slightly different there - time zones tend to change much > less frequently than provider pricings so publishing a new library package > every now and then is probably sufficient. > > On Sat, Jul 11, 2020 at 2:38 PM Samuel Marks <samuelma...@gmail.com> > wrote: > >> The other solution is to create a new git repository just for frequently >> updated files like this one… I mean we don't want to end up like pytz do >> we? >> >> PS: A good thing about pytz is other languages literally just parse pytz's >> list for their own timezone implementation. No Python. Easy! - With this >> being in JSON, I could imagine using Terraform libraries in Go; instead of >> Libcloud; to do multicloud, and use this costing system to say where and >> when. >> >> Samuel Marks >> Charity <https://sydneyscientific.org> | consultancy <https://offscale.io >> > >> | open-source <https://github.com/offscale> | LinkedIn >> <https://linkedin.com/in/samuelmarks> >> >> >> On Thu, Jul 2, 2020 at 9:51 PM Jay Rolette <role...@infinite.io> wrote: >> >> > Same here! >> > >> > Thanks, >> > Jay >> > >> > On Wed, Jul 1, 2020 at 12:45 PM Francisco Ros <fj...@doalitic.com> >> wrote: >> > >> > > Hey Tomaz, >> > > >> > > I'd really love to see this :-) >> > > >> > > Thanks, >> > > Francisco >> > > >> > > > El 1 jul 2020, a las 12:00, Tomaz Muraus <to...@apache.org> >> escribió: >> > > > >> > > > Recently one of the Libcloud contributors (Eis-D-Z) published >> various >> > > > improvements to our price scraping scripts and added some new ones - >> > > > https://github.com/apache/libcloud/pulls/Eis-D-Z. >> > > > >> > > > I think it would now make sense to run those scraping scripts on a >> > > > continuous basis as part of our CI (e.g. once a day) and publish the >> > > > generated file to some well known location (e.g. public read-only S3 >> > > > bucket). >> > > > >> > > > In fact, that was also the plan when we originally >> > > > added libcloud.pricing.download_pricing_file function and related >> > > > functionality quite a long time ago. >> > > > >> > > > IIRC, the plan was to include an auto-generated pricing file >> directly >> > > > inside the git repo, but this is more complicated and I would need >> to >> > > > contact the ASF infra team if they even allow something like that >> > > (updating >> > > > and committing a change as a bot user on our CI - Travis CI). >> > > > >> > > > So for now, I will probably just publish this auto-generated >> > pricing.json >> > > > file to a public read-only S3 bucket (I will make sure to set up >> > correct >> > > > rate limits and alerts to prevent abuse, even though the pricing >> file >> > > > itself is quite small). >> > > > >> > > > What do other people think? >> > > >> > > >> > >> >