Locking the Dolmen to a dev commit #893
Conversation
|
Well we do need a Dolmen version that is compatible with 4.08 though, Gbury/dolmen#195 |
There was a problem hiding this comment.
Actually now that I think about it -- I think it would be better to do this in the PR that will solve #892 so that we can track (using git log / git blame) the actual reason why we bump the dependencies.
I know. I said I checked it was still the case after this commit. |
Actually this commit was the very first commit of my PR for MaxSMT. I prefer separate it and create an remainder issue on this PR to remind we have to revert the lock before releasing. |
Ah, sorry — I parsed "bv2nat and int2bv don't share their cache as before." as stating that |
My sentence was very misleading ;) |
|
I prefer waiting the fix for Dolmen as we will need to change the lock again. |
This commit locks the Dolmen used by AE to a specific commit in order to use the unreleased feature for custom statements. I also notice that the `with_cache` in `d_cnf` was silently shadow by the same function from the module `Expr` of `Dolmen`. The signature of this latter changed recently. I fix it by using the `with_cache` from Dolmen. Notice that the hash table of the cache is hidden in the closure of `with_cache` and `bv2nat` and `int2bv` don't share their cache as before.
Halbaroth
force-pushed
the
lock-dolmen-dev
branch
from
a89b7c9 to
ed9b53b
Compare
After OCamlPro#893 the version of dolmen in the sources.json no longer works. Also bumps the pinned nixpkgs.
After #893 the version of dolmen in the sources.json no longer works. Also bumps the pinned nixpkgs.
It seems the lock file haven't been updated correctly in the PR OCamlPro#893. We have to lock `dolmen` to the dev version.
It seems the lock file haven't been updated correctly in the PR OCamlPro#893. We have to lock `dolmen` to the dev version.
It seems the lock file haven't been updated correctly in the PR #893. We have to lock `dolmen` to the dev version.
This commit locks the Dolmen used by AE to a specific commit in order to use the unreleased feature for custom statements (see #892).
I also notice that the
with_cacheind_cnfwas silently shadow by the same function from the moduleExprofDolmen. The signature of this latter changed recently. I fix it by using thewith_cachefrom Dolmen. Notice that the hash table of the cache is hidden in the closure ofwith_cacheandbv2natandint2bvdon't share their cache as before.