Closed mzabaluev closed 2 years ago
Besides, is there a good reason to keep both time
0.3.x and chrono
in the dependency tree?
chrono
is used in tiny-http
, pretty much entirely to parse and emit timestamps in RFC1123 format:
I'd accept either a PR or even a concrete suggestion for simplifying these uses. Should time
be preferred over chrono
, the inverse? Both would be suitable for the parse / emit workload and since we don't expose the types directly in the API it should be a minor change.
Should
time
be preferred overchrono
, the inverse?
That depends on your usage of course, but I find chrono
's treatment of leap seconds to be troublesome.
chrono
is used intiny-http
, pretty much entirely to parse and emit timestamps in RFC1123 format
Right; time
does not provide that out of the box, and I don't know if it can be easily supported by the format description syntax in time
.
Regarding the pinning issue though, would it be OK to bump this crate's MSRV to 1.51 and unpin time
, and at what release would that be made available?
Regarding the pinning issue though, would it be OK to bump this crate's MSRV to 1.51 and unpin
time
, and at what release would that be made available?
This would be acceptable I think, I need to cut a release to respond to the Brotli vulnerability anyway so I can roll this change into that release.
@mzabaluev this is complete and released as v3.5.0 34d31346d3b6aa241baf756a23f980e76222fa56.
(Or rather, it will be, once crates.io stops 503ing when you try and publish a release.)
The pin for
time =0.3.2
, while helping you keep consistency on MSRV, unfortunately gets in the way of other crates that would like to depend ontime ^0.3.5
forchecked_add
/checked_sub
(which should really be used everywhere in preference over the panicky overloaded operators), but don't care that much about keeping their MSRV back.