Prepare the 2.14.0.dev1 release.

[jsirois]

[ci skip-rust]

[jsirois]

Internal (put these in a PR comment for review, not the release notes)

• Fix JVM symbol mapping debug output. (#16069)

• Staunch CI bleeding from missing digest on main and 2.13.x branch #15995. (#16064)

• Allow deducing flag name from option attribute name (#16040)

• Remove unnecessary pex_binary targets (#16025)

• [internal] remove the instance name option from config (#16026)

• A better solution for skipping CI steps for docs-only changes. (#16053)

• Skip CI checks on docs-only PRs. (#16041)

• Fix BUILD file edit from Upgrade Pex to 2.1.94. #16029. (#16049)

• Upgrade Rust to 1.62.0. (#16039)

• Update 2.13.x changelog with recent cherry-picks for 2.13.0a1 (#16024)

• Work around a Rust build breakage. (#16019)

• Improvements to CI job generation. (#15989)

• Prepare 2.13.0a1. (#16011)

• Enrich MissingDigest errors when we fail to recover from them (#16008)

• [internal] test batch and stream API paths for load/store from remote store (#15996)

• [internal] terraform: use immutable input for Terraform tool (#15037)

• [internal] grpc_util: add layer to capture network-level timing data (#15978)

• Prepare 2.12.0. (#15982)

• Prepare 2.11.1. (#15985)

• [internal] use CAS BatchReadBlobs API for small blob reads (#15969)

• [internal] add observation metric for REAPI GetActionResult calls (#15967)

• Remove 2.13.0a0 entries from 2.14.0.dev0 release notes (#15955)

• Add codec for ExternalTool known version field (#15950)

• Improve debugpy's capabilities (#15946)

---

[jsirois]

I can repro the CI failures and the crazy thing is they bisect to this commit, which touches no source code. Working on it but definitely stumped at the moment.

[benjyw]

I can repro the CI failures and the crazy thing is they bisect to this commit, which touches no source code. Working on it but definitely stumped at the moment.

That is bizarre!

[benjyw]

FWIW I can repro this on this branch on macos

[benjyw]

And not on the prior commit. So, yeah. Weird as hell.

[jsirois]

Ah, somehow missed this:

pants/src/python/pants/engine/internals/specs_rules.py

Lines 468 to 493 in 27628c3

	if field_set_to_default_to is PexBinaryFieldSet:
	warn_or_error(
	"2.14.0dev1",
	"referring to a `pex_binary` by using the filename specified in `entry_point`",
	softwrap(
	"""
	In Pants 2.14, a `pex_binary` can no longer be referred to by the filename that
	the `entry_point` field uses.
	This is due to a change in Pants 2.13, which allows you to use the `run` goal
	directly on a `python_source` target without requiring a `pex_binary`. As a
	consequence the ability to refer to the `pex_binary` via its `entry_point` is
	being removed, as otherwise it would be ambiguous which target to use.
	Note that because of this change you are able to remove any `pex_binary` targets
	you have declared just to support the `run` goal
	(usually these are developer scripts), as using `run` on the `python_source` will
	have the equivalent behavior.
	To fix this deprecation, you can use the `pex_binary`'s address to refer to
	the `pex_binary`, or set the `[GLOBAL].use_deprecated_pex_binary_run_semantics`
	option to `false` (which, among other things, will have `run` on a Python
	filename run the `python_source`).
	"""
	),
	)

I'll get a PR out for that deprecation.

[benjyw]

You've probably already noticed this, but have confirmed that this is due to the change in VERSION, not the other files.

Not sure why though.

[benjyw]

Changing the VERSION to, say, 2.13.99.dev99 passes, but 2.14.X.devY for any X, Y I've tried fails.

So, hypothesis: this has something to do with deprecations.

[benjyw]

The only deprecation I can find that seems relevant is

pants/src/python/pants/base/build_environment.py

Line 215 in 44167c8

start_version="2.14.0.dev0",

, and bumping it didn't help, and neither did removing that deprecation entirely.

[jsirois]

Thanks for digging @benjyw. The fix is in #16054. I'll re-base this PR and add the changelog entry as soon as that's in.

[thejcannon]

Ah this is my bad, too. #16056 should fix this for you and I think #16054 can be closed.

[jsirois]

This should go green, but trying to get it landed during working hours after the long weekend in Canada and in the US is going to be ~impossible. Lots of stuff is landing. I'll update once more with new lands this evening and then finish this out then.