bisect is not supported for OCaml 4.06

[nekketsuuu]

The opam file says that batteries needs bisect for testing. However, bisect is not supported for OCaml 4.06 (ocaml/opam-repository#10751).

In fact, opam install -t batteries failed under OCaml 4.06.0 whereas opam install batteries succeeded.

$ opam switch show
4.06.0
$ opam reinstall -t batteries
The following dependencies couldn't be met:
  - batteries -> bisect
Your request can't be satisfied:
  - bisect is not available because your system doesn't comply with ocaml-version < "4.06.0".

No solution found, exiting

[gasche]

Thanks for the report!

We should port bisect to 4.06, migrate to ppx_bisect, or disable coverage testing, whichever is easier.

[rixed]

FWIW: http://bugs.x9c.fr/view.php?id=140

[UnixJunkie]

I guess 'opam reinstall batteries' would work.
I guess the -t flag forces bisect to be required.

[UnixJunkie]

@xclerc

[gasche]

This was a two-lines change: ocaml/opam-repository#11472

[nekketsuuu]

Thanks to gasche's PR, now opam reinstall -t batteries works on OCaml 4.06.0. Thank you all :)

[rixed]

Given that:

1. the source code for bisect is not easy to patch (using darc on a private repo with no quick way to submit patches),
2. xclerc does not seem to respond to github or even his own Mantis bugtracker fast enough for us (at least these days),
3. I believe no one ever paid attention to any code coverage tool

I suggest this dependency be dropped and will be happy to provide a patch doing just that if everyone agrees.

[gasche]

Yes, I'm fine with this as well. I did the work of implementing bisect support in Batteries, but we never really started to use coverage information (I think the constraint that any changed or added code must come with tests is enough in practice), so dropping bisect support is fine. If anyone was amused to re-introduce it (it was not a lot of work in the first place), of course bisect_ppx would be a wiser choice today.

[UnixJunkie]

I'm also in favor of cutting dependencies and simplifying our build system. Great suggestion.

…

On 26 Feb 2018, at 18:49, Gabriel Scherer ***@***.***> wrote: Yes, I'm fine with this as well. I did the work of implementing bisect support in Batteries, but we never really started to use coverage information (I think the constraint that any changed or added code must come with tests is enough in practice), so dropping bisect support is fine. If anyone was amused to re-introduce it (it was not a lot of work in the first place), of course bisect_ppx would be a wiser choice today. ― You are receiving this because you commented. Reply to this email directly, view it on GitHub, or mute the thread.