Offer “did you mean” for assignment ops #10262

som-snytt · 2023-01-11T21:04:39Z

Backport the latest Dotty research on recommending members of nearly identical spelling. (MONIS.)

Bumps the edit distance to consider, sorts the result, and presents up to 4 suggestions. Those that are not first-rank receive a qualifying, "or perhaps?" An ellipsis indicates when there are other candidates not presented to the user.

Avoid suggesting wait for priv or name (that is more of a Wordle move). The tweaked restriction is that the edit distance cannot exceed half the word length (for either user identifier or candidate), but if all the chars written by the user appear in the candidate string, then use it, since swaps and repetitions are common typos. For example, for acb suggest abc but not for ade. (Tempted to take ade because of the common prefix.)

Sample dotty:

scala> c.priv
-- [E008] Not Found Error: ---------------------------------------------------------------------------------------------
1 |c.priv
  |^^^^^^
  |value priv is not a member of C - did you mean c.wait?
1 error found

Avoid offering advice for setter, as in this dotty message:

1 |def v = c y_= 1
  |        ^^^^^
  |        value y_= is not a member of C - did you mean c.x_=?

but do advise on ++=.

It is slightly incongruous to suggest did you mean +++ or x_=? probably because the shared element is the suffix _= or just =. Maybe edit distance should be 1 for the shortest identifiers. More precisely, when comparing two setters, strip the suffix first. Then normal criterion says don't recommend x for y, which would be tedious.

Dotty offers only one suggestion:

scala> "xyz".lengt
-- [E008] Not Found Error: ---------------------------------------------------------------------------------------------
1 |"xyz".lengt
  |^^^^^^^^^^^
  |value lengt is not a member of String - did you mean ("xyz" : String).length?
1 error found

scala> "xyz".lgnet
-- [E008] Not Found Error: ---------------------------------------------------------------------------------------------
1 |"xyz".lgnet
  |^^^^^^^^^^^
  |value lgnet is not a member of String - did you mean ("xyz" : String).lines?
1 error found

now we exclude lines in the first case (edit distance compared to token length) but include length in the second (because it is a word scramble):

scala> "xyz".lengt
             ^
       error: value lengt is not a member of String
       did you mean length?

scala> "xyz".lgnet
             ^
       error: value lgnet is not a member of String
       did you mean lines? or perhaps length?

SethTisue · 2023-01-21T04:53:13Z

LGTM but I'm wondering about 90c1fa8, what's that about?

som-snytt · 2023-01-21T05:52:08Z

@SethTisue https://discord.com/channels/632150470000902164/632628489719382036/1063609622113562724

I haven't tested javac on this score yet. I noticed scalac files 2> errs will announce 19 errors on stdout. It is more obvious for all error output to go to stderr.

But what about warnings? Do they also print to stderr? I hope so. Yes, same for warnings.

➜  snips scalac -Xlint warn-only.scala 2> errs
1 warning

I vaguely recall considering this question, perhaps before the reporters re-refactor, but I didn't go that far back to check. This represents my current thinking.

Dotty uses only stderr, and I opened a ticket to beg them to reconsider.

som-snytt · 2023-01-21T06:01:58Z

I had to think too hard about what javac might warn about.

➜  snips javac -Xlint JWarn.java 2> errs
➜  snips cat errs
JWarn.java:5: warning: [empty] empty statement after if
                        ;
                        ^
JWarn.java:6: warning: [deprecation] getId() in Thread has been deprecated
                System.out.println(Thread.currentThread().getId());
                                                         ^
2 warnings

BTW scalac could include both the lint name and the -Wconf category cat name in the message.

SethTisue · 2023-01-21T14:50:44Z

Could 90c1fa8 go in a separate PR, or is there some necessary connection to the rest of the changes here?

som-snytt · 2023-01-21T15:20:57Z

Yes, it could.

SethTisue

LGTM, but let's wait to merge until scala/scala-dev#827 has been addressed so we can get a go-ahead from Jenkins

som-snytt · 2023-01-24T20:08:13Z

Squashed using artisanal squashing techniques.

som-snytt · 2023-01-25T20:54:35Z

Oh, I needed this today! Except...

➜  scala git:(tweak/not-member-help) ✗ scala
Welcome to Scala 2.13.10 (OpenJDK 64-Bit Server VM, Java 19).
Type in expressions for evaluation. Or try :help.

scala> import collection.immutable.LazyList
import collection.immutable.LazyList

scala> 42 :: LazyList.empty
          ^
       error: value :: is not a member of scala.collection.immutable.LazyList[A]
       did you mean ++: or :++?

scala>
:quit
➜  scala git:(tweak/not-member-help) ✗ skala
Welcome to Scala 2.13.11-20230124-195821-e781990 (OpenJDK 64-Bit Server VM, Java 19).
Type in expressions for evaluation. Or try :help.

scala> import collection.immutable.LazyList
import collection.immutable.LazyList

scala> 42 :: LazyList.empty
          ^
       error: value :: is not a member of scala.collection.immutable.LazyList[A]
       did you mean +:, /:, :+, or :\? or...?

oh man, let me filter out those deprecations.

did you mean +: or :+? or perhaps ++: or :++?

d'oh! of course it doesn't do conversions yet.

scala> 42 :: LazyList.empty
          ^
       error: value :: is not a member of scala.collection.immutable.LazyList[A]
       did you mean #::, +:, or :+? or perhaps #:::? or...?

The full list of alternatives to present in this case

ALTS[List((1,#::), (1,+:), (1,:+), (2,#:::), (2,++:), (2,:++))]

So the final wistful or...? is an intimation of the last two results, which happen to be the two low-quality results in 2.13.10.

SethTisue · 2023-01-25T21:03:22Z

so did you want to wait to merge this until you add some of those improvements, or should we book the progress?

som-snytt · 2023-01-25T21:27:06Z

I'll add the commit in just a second. I just excluded any2stringadd from the results...

som-snytt · 2023-01-25T22:13:56Z

That second commit was an exceptionally low-throughput slog. "Just give me all the available implicits" and "I just want to exclude the arrow of pairing." Hey, Cupid's arrow is also for pairing. This Valentine's, we should celebrate tuples instead of couples.

som-snytt · 2023-01-26T00:22:04Z

The run/analyzerPlugins.scala test showed extra calls because expressions such as i += 1 induce a failure (because int has no += member). The third commit adds DeferredTypeError to capture the mechanism and crank the handle only if the string message is requested. This may be safe for short-term deferral, in this case, as it decides whether the expression can be rewritten or just fail with the error message. It would not be appropriate for long-lived warnings which are buffered indefinitely.

Backport dotty behavior of ranking results. Improve by offering a few alternatives, with subtle wording to differentiate first rank from the rest. Allow shorter matches if it looks like the typo is char swaps and repetitions.

som-snytt · 2023-01-29T16:39:07Z

@SethTisue I dropped the commit for "conversions help" until I review it further (for performance reasons). Rebased just the commit you approved.

SethTisue · 2023-01-31T02:12:03Z

One small step for a man, one giant leap for the Oxford comma.

scala-jenkins added this to the 2.13.11 milestone Jan 11, 2023

som-snytt marked this pull request as ready for review January 12, 2023 01:40

som-snytt marked this pull request as draft January 12, 2023 17:08

som-snytt changed the title ~~Offer did you mean for assignment ops~~ Offer did you mean for assignment ops [ci: last-only] Jan 14, 2023

som-snytt force-pushed the tweak/not-member-help branch 3 times, most recently from fb6b4b4 to 911eccc Compare January 15, 2023 03:09

som-snytt marked this pull request as ready for review January 15, 2023 07:13

lrytz requested a review from SethTisue January 16, 2023 09:30

som-snytt force-pushed the tweak/not-member-help branch from 4f9ba73 to 7ae8c6b Compare January 18, 2023 04:57

som-snytt mentioned this pull request Jan 21, 2023

Report total errors on stderr #10272

Merged

som-snytt force-pushed the tweak/not-member-help branch from 7ae8c6b to 6858d20 Compare January 21, 2023 15:20

SethTisue approved these changes Jan 22, 2023

View reviewed changes

som-snytt force-pushed the tweak/not-member-help branch from 6858d20 to e781990 Compare January 24, 2023 20:06

som-snytt force-pushed the tweak/not-member-help branch from e781990 to bc77b62 Compare January 25, 2023 22:05

som-snytt force-pushed the tweak/not-member-help branch from bc77b62 to 538abc8 Compare January 26, 2023 00:16

Improve 'did you mean?' suggestions

bb82366

Backport dotty behavior of ranking results. Improve by offering a few alternatives, with subtle wording to differentiate first rank from the rest. Allow shorter matches if it looks like the typo is char swaps and repetitions.

som-snytt force-pushed the tweak/not-member-help branch from 538abc8 to bb82366 Compare January 29, 2023 16:36

som-snytt changed the title ~~Offer did you mean for assignment ops [ci: last-only]~~ Offer did you mean for assignment ops Jan 29, 2023

SethTisue changed the title ~~Offer did you mean for assignment ops~~ Offer “did you mean” for assignment ops Jan 31, 2023

SethTisue merged commit 2686321 into scala:2.13.x Jan 31, 2023

som-snytt deleted the tweak/not-member-help branch January 31, 2023 06:17

som-snytt mentioned this pull request Mar 7, 2023

Ask did you mean only for accessible members. scala/scala3#17067

Closed

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

Offer “did you mean” for assignment ops #10262

Offer “did you mean” for assignment ops #10262

som-snytt commented Jan 11, 2023 •

edited

Loading

SethTisue commented Jan 21, 2023

som-snytt commented Jan 21, 2023 •

edited

Loading

som-snytt commented Jan 21, 2023 •

edited

Loading

SethTisue commented Jan 21, 2023

som-snytt commented Jan 21, 2023

SethTisue left a comment

som-snytt commented Jan 24, 2023

som-snytt commented Jan 25, 2023

SethTisue commented Jan 25, 2023

som-snytt commented Jan 25, 2023

som-snytt commented Jan 25, 2023

som-snytt commented Jan 26, 2023

som-snytt commented Jan 29, 2023

SethTisue commented Jan 31, 2023

Offer “did you mean” for assignment ops #10262

Offer “did you mean” for assignment ops #10262

Conversation

som-snytt commented Jan 11, 2023 • edited Loading

SethTisue commented Jan 21, 2023

som-snytt commented Jan 21, 2023 • edited Loading

som-snytt commented Jan 21, 2023 • edited Loading

SethTisue commented Jan 21, 2023

som-snytt commented Jan 21, 2023

SethTisue left a comment

Choose a reason for hiding this comment

som-snytt commented Jan 24, 2023

som-snytt commented Jan 25, 2023

SethTisue commented Jan 25, 2023

som-snytt commented Jan 25, 2023

som-snytt commented Jan 25, 2023

som-snytt commented Jan 26, 2023

som-snytt commented Jan 29, 2023

SethTisue commented Jan 31, 2023

som-snytt commented Jan 11, 2023 •

edited

Loading

som-snytt commented Jan 21, 2023 •

edited

Loading

som-snytt commented Jan 21, 2023 •

edited

Loading