New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
Release Agda 2.6.1 #4324
Comments
Thanks for producing the new version of Agda. My repository https://github.com/martinescardo/TypeTopology doesn't type check anymore. I haven't had an opportunity to investigate. But running "agda AllModulesIndex.lagda" on "source/" gives
on the module UF-Yoneda. My worst fear, in every release, is that Agda will infer types in a different way... |
My repository https://github.com/martinescardo/HoTT-UF-Agda-Lecture-Notes works fine with the new version, thankfully. |
There is only one thing that goes wrong in 37137 lines of code, at UF-Yoneda.lagda line 581. The code is
If we supply the implicit arguments as
then the whole repository type checks. It is the fifth implicit argument |
I think one way to spot this incompatibility before each release is set up
a testing running regularly on GitHub and report the problem once the
regression is introduced.
I can write a template script for running testing in due course...
|
@martinescardo that looks like it could be an instance of #3960. I've spent quite some time trying to debug that one, but unfortunately all my attempts so far have failed. A small example where Agda 2.6.0.1 can infer all the implicits but the current master doesn't would be really useful. |
I see. This same error has already been reported there. I will try to shrink the code giving the error. |
Here is the requested, small, self-contained example:
You need to uncomment the above line to be able to type check this with the release candidate. |
Compilation seems to require quite a bit more memory, compared to the previous release. I guess it is for this reason that the makefile now (by default) gives the RTS option
Can we address this problem in some way? Should we add this RTS option to the Cabal file? |
No more open issues for 2.6.1 🎉. @asr can you prepare a new release candidate? |
Yes. I'll prepare the second release candidate. |
We could at least document the issue. |
Was not that RTS option added in the Lines 709 to 713 in 204183c
?
Is not the issue documented by Lines 27 to 33 in 204183c
? |
In the above commit, I updated the list of closed issues in the CHANGELOG. Anything missing or misplaced? |
Looks good to me. |
The current master (af74e45) is in good shape for preparing the second the release candidate, so the |
* Removed `-Werror`. * Removed `-fprof-auto`. * Removed Cabal test-suite. * Added `doc/user-manual.pdf`.
I uploaded to Hackage the second release candidate Agda-2.6.0.1.20200307 and I tested its installation on Ubuntu with
Could someone test the installation on macOS and Windows using the following instructions: $ cabal get https://hackage.haskell.org/package/Agda-2.6.0.1.20200307/candidate/Agda-2.6.0.1.20200307.tar.gz
$ cd Agda-2.6.0.1.20200307
$ cabal install |
I wasn't thinking in the release. |
I started the preparation for the release. |
I'm getting the error
when I tried to update the submodule commit of the standard library. I guess the error is due to the changes in commit agda/agda-stdlib@ef78949. Could someone push a fix into this branch please. You can reproduce the error on the above branch by running
|
@gallais, thanks for fixing the problem. |
I agree. Could you document the issue (in the |
I just saw that @Soares reported long build times: "The Agda dev version has been building on my machine for the last 2h, with no sign of stopping". @Soares, can you identify the problem?
|
I installed the prereqs listed here and then ran After 172 min, the build failed on package ~320 out of ~360 (IIRC), having exhausted the heap. I restarted the build, hoping (correctly) that it saved the first ~320 packages; the build then completed in 5min & succeeded.
I'm not sure. It only took one (of four) CPUs, and that CPU was pegged at 100% for the duration. I had
Nope, just a vanilla Other facts: machine is a 2019 Thinkpad, so a laptop, but a reasonably meaty one. OS Ubuntu 18-point-something, and I think the Haskell version was just whatever-apt-gives-you but I'm not sure. (Can't be much more specific, sorry; I no longer have access to that machine.) Reading back through the thread, I'm pretty sure this is the same thing @nad noted in this comment. |
This includes
I'm guessing that this includes time for installing other Haskell packages as well, not just Agda. |
Then I'll prepare the release from the current |
* Removed `-Werror`. * Removed `-fprof-auto`. * Removed Cabal test-suite. * Added `doc/user-manual.pdf`.
Agda 2.6.1 has been released. Good job everyone! |
Hooray! Thanks for your hard work! |
@Soares, how much memory has the laptop? |
@asr 16G mem + 16G swap |
After the release of the first candidate, let's continue the discussion here.
The text was updated successfully, but these errors were encountered: