fitzgen (he/him) (Feb 13 2026 at 18:02): fitzgen (he/him) (Feb 13 2026 at 18:02):As the OOM-handling effort progresses, it has become clear that we probably shouldn't name our OOM-handling collections the same as their equivalent abort-on-OOM alloc/std types. This is because things get pretty confusing during this period of time where we haven't migrated everything over to the OOM-handling collections (and almost certainly won't have migrated 100% any time soon; we may never even bother migrating the stuff under wasmtime::compile::*) and so when you see a Vec you have to wonder "is this abort-on-OOM alloc::vec::Vec or OOM-handling wasmtime_core::alloc::Vec?" That is annoying and we shouldn't add the additional and unnecessary mental load on ourselves.
So that begs the question: what naming scheme should we use for our OOM-handling, fallible-allocation-only collections?
The following candidates will use Vec as an example, but to be clear this will naming scheme also apply to our OOM-handling String, HashMap, etc...
Some initial ideas from @Alex Crichton :
TryVec- perhaps accurate, but long.TryIndexMapis a lot.TVec- personally I kinda like the ring of "letter plus collection". It's short and clearly "notstd"WVec- wasmtime's vec?XVec- any other random letterOomVec- I don't like this b/c the container handles OOMs, it doesn't make OOMsVecT(orVecWor such) - subjectively not a fan
Some more ideas:
WtVec -- another riff on "wasmtime's vec", kinda fun to read as "what the vec"OomHandlingVecOhVec -- shortened version of "OOM-handling"FallibleVecFVec -- f for fallibleQVec -- q for "question mark" which you use to propagate OOM errorsXec/XashMap/Xtringetc -- very xtreme, also cool cyberpunk (or post-cyberpunk?) transhumanism vibesAfVec - allocation fallibility vec, cool afZVec -- z is a rare and cool letterI hate every one of these options, pls help
Alex Crichton (Feb 13 2026 at 18:07):I hate every one of these options, pls help
honestly I agree with this take, but I'd also say that my dissatisfaction with these names isn't enough to overcome the desire to rename and avoid name clashes with the standard types
fitzgen (he/him) (Feb 13 2026 at 18:07):oh for sure, I think we are in the same boat here
Alex Crichton (Feb 13 2026 at 18:08):OpVec - "oom propagating" or "overpowered"
fitzgen (he/him) (Feb 13 2026 at 18:08):the original pusher
Alex Crichton (Feb 13 2026 at 18:09):I'd ever-so-slightly lean towards TVec still so far along the lines of "T for try" where the libstd methods are try_* so far
fitzgen (he/him) (Feb 13 2026 at 18:12):I don't know why by TVec just feels weird to me. maybe because T is so often a type param
I think the least-worst choices IMO rn are
TryVecQVecXec (it is so ridiculous, but also it is kinda growing on me and it is very concise) Alex Crichton (Feb 13 2026 at 18:12):I also didn't recoil as much as I thought I would at Xec
fitzgen (he/him) (Feb 13 2026 at 18:13):lol we could also do Wec, WashMap, etc and if you use a polish pronunciation of "W" then it is the same as "vec"
Alex Crichton (Feb 13 2026 at 18:13):I'm feeling like leaning towards something small to demarcate these collections insofar as I'd rule out TryVec
Alex Crichton (Feb 13 2026 at 18:13):lol well the polish part doesn't work soo well for Wtring
fitzgen (he/him) (Feb 13 2026 at 18:13):it really doesn't
Alex Crichton (Feb 13 2026 at 18:14):I think I might still lean more towards a prefix though
Alex Crichton (Feb 13 2026 at 18:14):and Q* seems fine by me
Alex Crichton (Feb 13 2026 at 18:14):QHashMap, QString, QVec, QIndexMap, ...
Alex Crichton (Feb 13 2026 at 18:14):QHashSet
Alex Crichton (Feb 13 2026 at 18:15):QPrimaryMap
Ramon Klass (Feb 13 2026 at 18:15):shouldn't wec come before xec after vec? (sorry)
Alex Crichton (Feb 13 2026 at 18:15):QEntitySet
fitzgen (he/him) (Feb 13 2026 at 18:15):XndexMap doesn't really work
fitzgen (he/him) (Feb 13 2026 at 18:15):looks like xanax
Alex Crichton (Feb 13 2026 at 18:15):lol
fitzgen (he/him) (Feb 13 2026 at 18:15):okay so QVec et al?
Alex Crichton (Feb 13 2026 at 18:16):my gut is _Vec, although I'm not wed to Q in particular there
fitzgen (he/him) (Feb 13 2026 at 18:16):Q is a p cool letter
Alex Crichton (Feb 13 2026 at 18:16):Q sort of makes me think "qt vec"
fitzgen (he/him) (Feb 13 2026 at 18:16):hrm
Alex Crichton (Feb 13 2026 at 18:17):Q feels not unreasonable to type though
fitzgen (he/him) (Feb 13 2026 at 18:17):"double yoo" is just too long
Alex Crichton (Feb 13 2026 at 18:17):good point
Alex Crichton (Feb 13 2026 at 18:17):gotta have a single-syllable-letter...
Alex Crichton (Feb 13 2026 at 18:17):although that literally only rules out W
Alex Crichton (Feb 13 2026 at 18:17):I'd say QVec
fitzgen (he/him) (Feb 13 2026 at 18:17):never thought about that before
fitzgen (he/him) (Feb 13 2026 at 18:18):okay lets go with Q
fitzgen (he/him) (Feb 13 2026 at 18:18):it has a nice ring to it
Chris Fallin (Feb 13 2026 at 19:15):I'm late to this convo but I'd second the "Q_ looks like Qt types" that could lead to confusion (that's certainly where my mind went). very bikeshed comment of me (but that's the thread, sorry!) but ZVec maybe? reminiscent of "zero" and embedded stuff somehow
Paul Osborne (Feb 13 2026 at 19:38):Having done my time with Qt, I had the same thought @Chris Fallin but I do think the form makes sense. I've seen S prefixes for "safe" variants but don't love that either as safe has a very defined meaning in Rust already.
Alex Crichton (Feb 13 2026 at 19:44):BVec for "bikeshed vec"
fitzgen (he/him) (Feb 13 2026 at 19:45):BBTreeMap
fitzgen (he/him) (Feb 13 2026 at 19:45):BsVec -- bikeshed vec
Alex Crichton (Feb 13 2026 at 19:46):I'd agree though "safe" wouldn't be right for this in that both variants (libstd and fallible) are safe Rust-wise
Alex Crichton (Feb 13 2026 at 19:46):I'm also loathe to acknowledge that Qt has a monopoly on Q-prefixed things in code...
Alex Crichton (Feb 13 2026 at 19:47):not that I've ever done anything with or in Qt myself heh
Alex Crichton (Feb 13 2026 at 19:47):KVec - "kernel suitable"
fitzgen (he/him) (Feb 13 2026 at 19:50):that seems decent to me
Chris Fallin (Feb 13 2026 at 20:00):my response is just... "K"
Chris Fallin (Feb 13 2026 at 20:00):(seems fine!)
Christof Petig (Feb 13 2026 at 20:51)::thinking: K is fine, the easily pronounceable Wa prefix has not been mentioned, yet ... WaVec reminds me more of Wasmtime than KVec.
Victor Adossi (Feb 14 2026 at 01:01):It sounds like KVec is already the winner here, but allow me to offer some more paint buckets:
CheckedVec/CzechedVec
FVec ("F" for fallible)
AllocVec/AVec
OFVec ("OF" for "OOM Free")
OomfVec (personal favorite)
matt rice (Feb 17 2026 at 18:51):I find the CzechedVechumorous, but would shorten it to CzVec,
Last updated: Feb 24 2026 at 04:36 UTC