[new release] atdgen-codec-runtime, atd, atdgen-runtime, atds, atdj and atdgen (2.2.0)

[rgrinberg]

Runtime for atdgen generated bucklescript converters

• Project page: https://github.com/ahrefs/atd

CHANGES:

• Add support for merging double annotations (<ocaml from="ProtoA"><ocaml predef>)

• Add tests for annotation merging and target-specific annotations

[kit-ty-kate]

The packages are failing to build:

#=== ERROR while compiling atd.2.2.0 ==========================================#
# context              2.0.6 | linux/x86_64 | ocaml-base-compiler.4.10.0 | pinned(https://github.com/ahrefs/atd/releases/download/2.2.0/atd-2.2.0.tbz)
# path                 ~/.opam/4.10/.opam-switch/build/atd.2.2.0
# command              ~/.opam/4.10/bin/dune build -p atd -j 72 @install
# exit-code            1
# env-file             ~/.opam/log/atd-23-9af213.env
# output-file          ~/.opam/log/atd-23-9af213.out
### output ###
# File "atdgen/test/dune", line 199, characters 40-51:
# 199 |  (deps    (:atd test_annot_error.atd) %{bin:atdgen})
#                                               ^^^^^^^^^^^
# Error: Program atdgen not found in the tree or in PATH
#  (context: default)

- File "atdgen/test/dune", line 199, characters 40-51:
- 199 |  (deps    (:atd test_annot_error.atd) %{bin:atdgen})
-                                               ^^^^^^^^^^^
- Error: Program atdgen not found in the tree or in PATH
-  (context: default)

[rgrinberg]

Hmm, I have no idea why dune is looking for the executable. It's only necessary for running the tests. In any case, I'm a bit short on time, so I cannot continue to investigate.

@pirbo do you want take over the releasee? Feel free to make whatever changes are necessary to make sure the tests aren't getting in the way. Please don't send patches to opam-repository however, just tag & release from upstream.

[kit-ty-kate]

from afar the error seems a bit like the one encountered in #15879 for rpc but with tests rules instead of documentation rules.

[rgrinberg]

Indeed, but in that issue the documentation was part of the package. I don’t see how I’ve made my tests part of the package in atd.

…

On Mar 9, 2020, 4:26 PM +0000, Kate ***@***.***>, wrote: from afar the error seems a bit like the one encountered here #15879 for rpc but with tests rules instead of documentation rules. — You are receiving this because you modified the open/close state. Reply to this email directly, view it on GitHub, or unsubscribe.

[pirbo]

All my patches needed to fix 2.1.0 have been upstreamed @XVilka.
By uptreaming them, I detected a new potential problem in 2.2.0 for which I gave my workaround in ahrefs/atd#207 (comment) but as it was more hallucinating what the test was supposed to be that fixing something obvious (and as I never ever used ppx-deriving-sexp myself), I didn't turn that into an PR.

The problem here seems to be again simply a missing (package ...) restriction in a runtest rule. I can have a quick look at this particular issue but sorry my hands are too full (as everybody else I know) to be involve in the general maintainance of atd.

[avsm]

@rgrinberg indicated that the atd release would happen after dune 2.5.0 is out, as there's a bugfix in there that requires the dune fix.

[pirbo]

Good and I'm happy to de-involve myself :-)
I still put my conclusion as I spent the last hour on that: It looked to me that simply removing %{bin:atdgen} from the deps of test_annot_error.stderr fixes the problem (and dune build @atdgen/runtest still works perfectly in a cleaned env as dune knows it need to build %{bin:atdgen} when it appears in the "rule").

[XVilka]

Just for the record - Dune 2.5.0 had been merged