@bitfield@MattJamesBoyle@_rsc Hi @bitfield, I think Go 1.22+ is actually the cutoff when Go 1.20 (for example) will stop building code from the future, and 1.22 was chosen as the cutoff because of the change to loop variable semantics that will likely occur then.
See for example:
https://t.co/xVpNSEnYzi
@NateTheFinch@rogpeppe@_rsc Hi @NateTheFinch, FWIW, the related language change proposal for updating `range` might be a good place to start if you haven't already read it:
https://t.co/t8g9W7ID14
@largedatabank@benbjohnson@felixge@eatonphil@dovunderscore Hi @largedatabank, maybe a short blog or tweet (again) or first toot about gcassert?
@eatonphil, FWIW, tests are also a common way to keep allocations at 0 for a chunk of Go code using some of the helpers from the testing package. Example from stdlib: https://t.co/fJpqzF72VY
@dvyukov In any event, I know Blender is fresh prototype, and seems it'd be very nice to have engines support EXECUTE(input')->(input'', coverage) as you described, but partly thinking about today with current engines...
@dvyukov The engine then is effectively encouraged to supply more input data to get more coverage, and it has the control to mutate that input.
(I've done that before, and it seemed to materially help vs. pseudo rand, including for example with thepudds/fzgen:
https://t.co/MFbxtVtV7N
@SIGTIM@grittygrease@_rsc Regarding Oblivious HTTP, is Cloudflare or someone else currently operating a proxy or planning to do so such that it could be used by random or large Open Source projects?
@grittygrease@SIGTIM@_rsc (Sorry, slightly shortened to make it fit... but see the link for the full text).
You might have a different take than that, but wanted to at least post the follow up...
@grittygrease@SIGTIM@_rsc "Another problem is that they are difficult to explain and inspect. The math is solid, and if you're collecting sensitive data, you need something like them. But if you're not sending anything sensitive, then it could be better to use a simpler system that is easier to explain"
@adamhjk@kelseyhightower There are some core uses that are popular, like handy new generic maps and slices packages, updates to atomics, or db access, but there hasn't been an ecosystem-wide explosion of "Rewrite it in Generics".
@LachlanEvenson@stefanprodan@cue_lang@HelmPack Hi @sabre1041, it seems the first step of a Helm proposal is emailing cncf-helm to gauge interest? Are those usually pretty brief?
@stefanprodan (or others), do you have a mental pic of how Helm support for CUE should look, or "if I could wave a magic wand I'd wish it solves X"?
@adamhjk@kotakmakan Easier to say - I’ll just build a compelling open source product, and figure out $$ later.
Thanks for the link to the book. Will read.
@adamhjk@kotakmakan A challenge of trying to emulate Red Hat is that it’s really not very easy to build an enterprise sales org, with a strong inside sales team, effective professional services team, etc., especially when you are starting out initially small.
@goLibHunt Minor clarification: I didn’t have anything to do with that blog post.
The reason I was mentioned in the post is they used an arenas benchmark that I had put together a year or so ago:
https://t.co/UOvrjgedpf