revamp walk/1 #2655

pkoppstein · 2023-07-05T06:01:31Z

Resolves #2584; also resolves #2611

Resolves jqlang#2584; see also jqlang#2611

itchyny · 2023-07-05T11:05:24Z

The output of {x:0} | walk(.,1) changes by this PR. Is it intended?

pkoppstein · 2023-07-05T12:06:09Z

It is not the primary intention, which was to make the definition in terms of map and map_values as simple as possible while ensuring correct behavior when the arg was "simple" (no top-level commas). The current behavior was unintended.

Perhaps the manual should mention this example?

itchyny · 2023-07-05T14:08:58Z

I just wanted to confirm that you're aware of the behavior change. Even if not mentioned in the manual, we need reasonable explanations to any behavior changes. Typical reason is that the original behavior was unintentional bug, but making the implementation simpler is also a good reason. This time, I think changing this behavior would be Okay. By the way, in gojq I use last to make it work like jq but I will drop it...

nicowilliams · 2023-07-05T21:51:07Z

Hmm, I think we want the original behavior, yeah.

nicowilliams · 2023-07-05T21:52:30Z

Or maybe I should read @2584 before commenting...

pkoppstein · 2023-07-06T07:16:52Z

@nicowilliams - Perhaps it would help to review the situation w.r.t. walk/1.

The old def of walk(f) was written without regard to the
possibility that f could be a sequence of expressions. I have not
double-checked but hopefully the official documentation
never came remotely close to specifying what the semantics should be
in such cases.
In the case in question, I think it's generally agreed that one would want

{x:0} | walk(.,1) to be equivalent to {x:0} | walk(.), walk(1)

In this and other cases, jaq and the new walk are in agreement,
and as you have no doubt noticed,  @itchyny indicated that he had to go out of his way
to align gojq's `walk` behavior with jq's.

The new def elegantly resolves several important documented issues; this consideration,
in conjunction with the fact that the new def of walk/1 is as close to a "declarative" specification of
the desired functionality as one could achieve or want in jq, seem to me very compelling.

These are stored in vars, so we need to make sure that nlocals is large enough to account for that

…ang#2737)

…jqlang#2754)

Close jqlang#1885, jqlang#2140, jqlang#2011, jqlang#2220, jqlang#2485, jqlang#2073 Rename the FORK_OPT opcode to TRY_BEGIN, add a TRY_END opcode, and wrap errors when raising through a TRY_END so that they will not be caught by the matching TRY_BEGIN. Now a `try exp catch handler` expression generates code like: TRY_BEGIN handler <exp> TRY_END JUMP past_handler handler: <handler> past_handler: ... On backtrack through TRY_BEGIN it just backtracks. If anything past the whole thing raises when <exp> produced a value, then the TRY_END will catch the error, wrap it in another, and backtrack. The TRY_BEGIN will see a wrapped error and then it will unwrap and re-raise the error. If <exp> raises, then TRY_BEGIN will catch the error and jump to the handler, but the TRY_BEGIN will not stack_save() in that case, so on raise/backtrack the TRY_BEGIN will not execute again (nor will the TRY_END).

…c.) (jqlang#2744) - Add error/0 and mentions null input behavior (close jqlang#2231) - Explain value iterator suffix syntax .foo[] (close jqlang#1047) - Mention array slicing is also zero-based (close jqlang#2094) - Add examples of input and inputs filters (close jqlang#2216, close jqlang#2470) - Improve sort_by about multiple values (close jqlang#2103, close jqlang#2467, close jqlang#2474) - Improve foreach section and simplify examples (close jqlang#1148, close jqlang#2169) - Fix recurse/1 document on how it is identical using recurse/2 (close jqlang#2036, close jqlang#2412) - Add non-string examples of index/1, rindex/1 (close jqlang#1422) - Simplify the example of truncate_stream/1 (close jqlang#1736)

Signed-off-by: David Korczynski <david@adalogics.com>

{ BINDING: ExpD } wasn't freeing BINDING. Fixes https://bugs.chromium.org/p/oss-fuzz/issues/detail?id=60865

Changes mentioned based on picking user facing changes from: git log --oneline -r master...jq-1.6 | grep -v Merge

correct grammar, add attributions, clarify abs

Backfill with references to PRs & issues in NEWS.md

We now have an official Discord server and most maintainers are hanging out there. It would be a good idea to redirect questions to Discord.

* Bump up Bootstrap to v5.3.1, Bootstrap Icon to v1.10.5. * Use autoComplete.js to drop dependency on jQuery and typeahead.js. * Support dark mode. * New svg logo and icon with responsive color mode support. * Normalize section ids to lower kebab-case for easiness of linking. * Use relative paths for links for local development (--root /output). * Various markup cleanups and accessibility improvements.

itchyny · 2023-07-31T10:47:50Z

src/builtin.jq

@@ -249,17 +249,18 @@ def bsearch($target):
      else .[2]
      end
  end;
-
+#


I prefer this line to be an empty line rather than an empty comment.

OK, but someone should probably run: s/^# *$//

itchyny · 2023-07-31T10:48:07Z

Please rebase this PR, thanks.

Resolves jqlang#2584; see also jqlang#2611

Note that according to the new def: `{x:0} | walk(.,1)` is equivalent to: {x:0} | walk(.), walk(1)

pkoppstein · 2023-07-31T11:38:54Z

@itchyny wrote:

Please rebase this PR, thanks.

I did, but somehow a "merge" was added as well. Feel free to "squash" as desired.

itchyny · 2023-07-31T11:46:36Z

Could you rebase the commit onto the latest remote master branch? This PR includes the commits ahead and I can't review the diff now.

Resolves jqlang#2584; also resolves jqlang#2611 and supersedes jqlang#2655 Note that according to the revised implementation: `{x:0} | walk(.,1)` is equivalent to `{x:0} | walk(.), walk(1)`

pkoppstein · 2023-07-31T12:34:38Z

Given the difficulty merging this PR, I'm closing it in favor of #2795

nicowilliams · 2023-07-31T15:15:06Z

@pkoppstein if you would only just use your terminal application and the git command you wouldn't have such difficulties and you wouldn't cause so many duplicate PRs.

pkoppstein · 2023-07-31T19:24:22Z

@nicowilliams wrote:

use your terminal application and the git command

In brief, I did, but itchyny said he could not use the result of a simple "Rebase" and "Push".

In long - well, let me know if and when you would like further details.

revamp walk/1

eeb08b5

Resolves jqlang#2584; see also jqlang#2611

pkoppstein requested a review from itchyny July 5, 2023 06:02

itchyny added performance bug labels Jul 5, 2023

wtlangford and others added 20 commits July 21, 2023 09:01

Increment nlocals when compiling STORE_GLOBAL ops (jqlang#2000)

5152602

These are stored in vars, so we need to make sure that nlocals is large enough to account for that

Add a regression test for jqlang#2000

99a77f7

Clarify the Docker image location (close jqlang#1864) (jqlang#2740)

7c2bf9b

manual.yml: some clarifications plus fixes for markdown bloopers (jql…

3553a9d

…ang#2737)

Remove AppVeyor configuration (jqlang#2741)

4732dd8

Fail quickly if not in git repository in scripts/version

3ed9289

Enable verbose output of make check

e0e1b22

Update AUTHOR for the next release

649d6b5

Add tests for --arg, --argjson, and similar options

c9c45d7

Validate manual schema

9044f54

Fix manual source code to follow the schema definition

34629ed

Add importlib_resources dependency for older Python (fix jqlang#2753) (…

5753722

…jqlang#2754)

Improve --help to mention all the command options

99a381b

Enable significand on macOS

123c780

Fix --stream-errors w/o --stream, and doc+test (close jqlang#2748)

f64ef1a

Add checksum signature of old release artifacts (close jqlang#1999)

dfdce8c

Fix stderr to output string with no decoration (fix jqlang#2063)

98f709d

Add compile fuzzer for OSS-Fuzz

cf3c11b

Signed-off-by: David Korczynski <david@adalogics.com>

itchyny and others added 17 commits July 27, 2023 17:01

Commit GPG signatures on release

4160a36

Fix release job to create a pull request for signatures

72f147e

Fix gh pr create option on release

b33725c

Fix PR creation by specifying the branch on release

5b9d075

Fix memory leak for { $foo: bar }

17889a1

{ BINDING: ExpD } wasn't freeing BINDING. Fixes https://bugs.chromium.org/p/oss-fuzz/issues/detail?id=60865

Fix PR creation by using fully qualified refname on release

d319eb2

Allow .[-1] in path expressions

086a156

Test negative indices in path expressions

336c3de

Replace NEWS with NEWS.md with more details and examples

28af007

Changes mentioned based on picking user facing changes from: git log --oneline -r master...jq-1.6 | grep -v Merge

NEWS.md: tweaks

70bbd10

correct grammar, add attributions, clarify abs

Backfill with references in NEWS.md (jqlang#2788)

ee2a215

Backfill with references to PRs & issues in NEWS.md

Fix typo in manual: "-seq" => "--seq"

a6eb055

Add JQ_FALLTHROUGH and use it to quiet warnings

c8b30df

Use -Wno-cast-function-type to quiet many warnings

f733a15

Redirect questions to Discord

82f7f77

We now have an official Discord server and most maintainers are hanging out there. It would be a good idea to redirect questions to Discord.

Update bug_report.md with Discord link

4af3f99

itchyny reviewed Jul 31, 2023

View reviewed changes

pkoppstein added 3 commits July 31, 2023 07:09

revamp walk/1

bdec9c0

Resolves jqlang#2584; see also jqlang#2611

revamp walk/1: rebase, add test cases

4665e81

Note that according to the new def: `{x:0} | walk(.,1)` is equivalent to: {x:0} | walk(.), walk(1)

Merge branch 'walk' of https://github.com/pkoppstein/jq into walk

9afda0a

pkoppstein mentioned this pull request Jul 31, 2023

Simple and efficient implementation of walk/1 #2795

Merged

pkoppstein closed this Jul 31, 2023

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

revamp walk/1 #2655

revamp walk/1 #2655

pkoppstein commented Jul 5, 2023 •

edited by itchyny

Loading

itchyny commented Jul 5, 2023

pkoppstein commented Jul 5, 2023

itchyny commented Jul 5, 2023

nicowilliams commented Jul 5, 2023

nicowilliams commented Jul 5, 2023

pkoppstein commented Jul 6, 2023

itchyny Jul 31, 2023

pkoppstein Jul 31, 2023

itchyny commented Jul 31, 2023

pkoppstein commented Jul 31, 2023

itchyny commented Jul 31, 2023

pkoppstein commented Jul 31, 2023

nicowilliams commented Jul 31, 2023

pkoppstein commented Jul 31, 2023

revamp walk/1 #2655

revamp walk/1 #2655

Conversation

pkoppstein commented Jul 5, 2023 • edited by itchyny Loading

itchyny commented Jul 5, 2023

pkoppstein commented Jul 5, 2023

itchyny commented Jul 5, 2023

nicowilliams commented Jul 5, 2023

nicowilliams commented Jul 5, 2023

pkoppstein commented Jul 6, 2023

itchyny Jul 31, 2023

Choose a reason for hiding this comment

pkoppstein Jul 31, 2023

Choose a reason for hiding this comment

itchyny commented Jul 31, 2023

pkoppstein commented Jul 31, 2023

itchyny commented Jul 31, 2023

pkoppstein commented Jul 31, 2023

nicowilliams commented Jul 31, 2023

pkoppstein commented Jul 31, 2023

pkoppstein commented Jul 5, 2023 •

edited by itchyny

Loading