[new release] elpi (1.17.0) #24098
[new release] elpi (1.17.0) #24098
Conversation
CHANGES:
Requires Menhir 20211230 and OCaml 4.08 or above.
Camlp5 8.0 or above is optional.
Compiler:
- Improve performance of separate compilation, in particular
extending a program with more clauses. This change may
break existing code which accumulates units containing
the spilling of a predicate *before* the unit declaring
the predicate signature.
Parser:
- Fix error message on unexpected keyword (was wrongly assuming the
keyword was `)` misleading the user)
Builtins:
- Change type of `declare_constraint` to `any -> any -> variadic any prop`
making it explicitly take at least two arguments
Trace browser:
- Fix elaboration of CHR rule with no condition
|
There are some CI failures running tests and
There are at least the following options to resolve this:
There are some other CI failures that are likely due to CI infra problem, but it is likely that they will either succeed or fail with the same problem listed above. |
|
But besides of the dependency problem when testing, CI reports suggests that the package is generally installable, so ↑ could be addressed later IMO. |
|
Lua Is not the cause of the failure, it is used to tally the results. Also, to my understanding, on ocaml 5 failures are unrelated. |
Ah, I see. So are lines like the below the actual hints about the failure? Since the error says "Error: Thread.create: Resource temporarily unavailable" I thought it might be due to some issue about the CI machine (as other platform were succeeding) I have tried to hit the "rebuild" on the CI job page a couple of times, but they ended up with the same place. I am not sure about the root cause, but it seems to be more related to the particular CI machine than the package and test cases so I don't see them as blocker for this PR.
Do you mean the following failing test cases: As the CI is not giving any information, I can't say anything. But it indeed does not look like a blocker to me. |
|
As the CI failures seem unrelated, I am in favor of considering for merge, but someone else should have a check. |
| "cmdliner" {with-test} | ||
| "dune" {>= "2.8.0"} | ||
| "conf-time" {with-test} | ||
| "atdgen" {>= "2.10.0"} |
There was a problem hiding this comment.
| "atdgen" {>= "2.10.0"} | |
| "conf-lua" {with-test} | |
| "conf-gnuplot" {with-test} | |
| "atdgen" {>= "2.10.0"} |
There was a problem hiding this comment.
As you already discussed this is not needed to make the tests succeed, so I don't have any strong opinion on adding it. Since the binaries are used to collect the test results, I wanted to point out that we have the necessary conf packages
There was a problem hiding this comment.
thanks, done upstream
|
I am rerunning the failing tests |
|
I suggest we merge it. If there are issues we can get bug reports from users on these platforms |
|
The issue with |
3 participants
ELPI - Embeddable λProlog Interpreter
CHANGES:
Requires Menhir 20211230 and OCaml 4.08 or above.
Camlp5 8.0 or above is optional.
Compiler:
extending a program with more clauses. This change may
break existing code which accumulates units containing
the spilling of a predicate before the unit declaring
the predicate signature.
Parser:
keyword was
)misleading the user)Builtins:
declare_constrainttoany -> any -> variadic any propmaking it explicitly take at least two arguments
Trace browser: