Update to mtime.2.0.0 #2166
Update to mtime.2.0.0 #2166
Conversation
patricoferris
force-pushed
the
mtime.2.0.0
branch
2 times, most recently
from
72d48af
to
08641f0
Compare
| @@ -88,7 +88,7 @@ let create_pp_seconds examples = | |||
| Float.neg_infinity examples | |||
| in | |||
| let finite_pp = | |||
| if absmax >= 60. then fun ppf v -> Mtime.Span.pp_float_s ppf v | |||
| if absmax >= 60. then fun ppf v -> Fmt.uint64_ns_span ppf (Int64.of_float v) | |||
There was a problem hiding this comment.
I kinda made this bit up to be honest, not sure what you would like to do here?
There was a problem hiding this comment.
Your choice seems reasonable to me since I think it is close to what was there with the previous version of mtime.
|
@patricoferris thanks for doing this set of PRs -- this change in mtime is causing quite the ripple! I feel conflicted about pushing the lower bounds to 2.0.0 because of the pressure it puts on dependencies, but we will have to if Eio keeps their change and we adopt it. Either way, looks like we have some upstream dependency shepherding to do before we can merge this. |
|
Thanks for taking a look @metanivek just wanted to bring it up for when the time comes (if it's a bit noisy feel free to close and just grab the diff at a later date ^^). I had originally started with a set of changes that were backward compatible but it would mean starting to distribute the internals of |
|
I'm happy to keep this open. Added the label just so we don't accidentally merge. 😅 One other point of consideration for us is our downstream consumers, like Tezos. We will need to coordinate this kind of change with them as well. This is definitely a lesson in the impact of breaking changes for "core" libraries in the ecosystem, like |
Co-authored-by: Kevin Smith <metanivek@gmail.com>
There was a problem hiding this comment.
Thanks for the initial work on this @patricoferris! I rearranged some code (mainly moving mtime extensions into irmin's import) so we don't have to sprinkle the conversions around.
…min-pack, irmin-pack-tools, irmin-mirage, irmin-mirage-graphql, irmin-mirage-git, irmin-http, irmin-graphql, irmin-git, irmin-fs, irmin-containers, irmin-cli, irmin-chunk and irmin-bench (3.8.0)
CHANGES:
### Added
- **irmin**
- Change behavior of `Irmin.Conf.key` to disallow duplicate key names by
default. Add `allow_duplicate` optional argument to override. (mirage/irmin#2252,
@metanivek)
- **irmin-pack**
- Add maximum memory as an alternative configuration option, `lru_max_memory`,
for setting LRU capacity. (@metanivek, mirage/irmin#2254)
### Changed
- **irmin**
- Lower bound for `mtime` is now `2.0.0` (mirage/irmin#2166, @patricoferris)
- **irmin-mirage-git**
- Lower bound for `mirage-kv` is now `6.0.0` (mirage/irmin#2256, @metanivek)
### Fixed
- **irmin-cli**
- Changed `--store irf` to `--store fs` to align the CLI with what is
published on the Irmin website (mirage/irmin#2243, @wyn)
A bit of a WIP PR for #2155 -- the main blockers outside of Irmin are in the Irmin-pack dependencies:
Happened to bump into this whilst trying to rebase the Eio port which now requires mtime >= 2