05:52 <steve_s> We discussed this with @fangerer: in theory, shouldn't the ABI use fixed width integers? The API can take the "regular" integers and cast/check the size depending on the real width of those?

06:26 <antocuni> steve_s: I don't understand the question

06:53 <steve_s> example: ctx_Long_FromLong would take int32_t, instead of long, because in theory long could be a different thing across compilers. HPyLong_FromLong could still take "long", but would have to handle the situation when long != int32_t somehow

06:54 <steve_s> I don't know if that's a real problem in practice..

07:51 <cfbolz> Random other hpy thing: I talked with Tim, I think it would be a good idea to have an hpy Twitter account to get a bit more visibility. If nobody is enthusiastic about doing it, I'm happy to do it in a low key way

08:21 <fangerer> cfbolz: since I'm not on Twitter, I've no objections if you create it. We just need to share credentials, right? Otherwise you will have to twitter all the time

08:22 <pmp-p> nothing on mastodon ?

08:30 <cfbolz> fangerer: yeah, we can, but how often do you expect to write something? If it's like twice a week I don't mind

08:33 <cfbolz> pmp-p: I'm not on mastodon, so I am not volunteering for that

08:38 <pmp-p> cfbolz: me not on twitter - so not sure - but i was thinking of some kind of bot, as i've seen a lot of "RT from xxxx (tweeter account)" on mastodon feeds

08:43 <fangerer> cfbolz: ok, let's see; we don't need to eagerly share credentials

08:44 <cfbolz> Is there a private repo for things like that yet?

11:36 <fangerer> not that I know of

11:44 <fangerer> Since it was again a topic during our meetup in Berlin: I wrote a summary about having a stack-allocated `HPyContext`. Please add your thoughts. https://github.com/hpyproject/hpy/issues/330