git/list[1] front-page[2] threads[3] people[4] search[5] about
 

Re: [PATCH 2/2] chainlint: reduce annotation noise-factor

From
Eric Sunshine <sunshine@sunshineco.com>
Date
Aug 29, 2024, 18:28 UTC
Message-ID
<CAPig+cTTXmaAjJrOOSKKDKvMAE+yD9wfoYii5C21jGpq=sqtyA@mail.gmail.com>
In-Reply-To
<ZtBHecRkFQkSAF6C@tanuki>
On Thu, Aug 29, 2024 at 6:03 AM Patrick Steinhardt <ps@pks.im> wrote:
Show 24 quoted lines
> On Thu, Aug 29, 2024 at 05:16:25AM -0400, Eric Sunshine wrote:
> > Note that the preceding change gave all problem annotations a uniform
> > "ERR" prefix which serves as a reasonably suitable replacement needle
> > when searching in a terminal, so loss of "?!" in the output should not
> > be overly problematic.
>
> Okay, now the "ERR" prefix becomes a bit more important because we drop
> the other punctuation. I'm still not much of a fan of it, though. Makes
> me wonder whether we want to take a clue from how compilers nowadays
> format this, e.g. by using "pointers".
>
> So this:
>     7   fish |
>     8   cow ?!AMP?!
>
> Would become this:
>     t/chainlint/pipe.actual:8: error: expected ampersands (&&)
>     7   fish |
>     8   cow
>             ^
>
> While this would be neat, I guess it would also be way more work than
> the current series you have posted. And whether that work is ultimately
> really worth it may be another question. Probably not.

Interestingly, I'm not always a fan of the sort of compiler output you suggest since I often have more difficulty interpreting the output and locating the actual problem[*] than if the annotation was merely inline, sitting immediately next to the problem itself.

Also, the vast majority of the time, chainlint will be flagging a missing "&&" at the end of line, so with the inline annotation, it's very easy to see (especially when colored) exactly where the problem is at a glance.

Hence, the cost of implementing "^" doesn't feel particularly worthwhile (and, with my limited Git time these days, I'm unlikely to do so).

[*] This is especially so when dealing with foreign code which is wider than my 80-column terminal or 80-column editor window, in which the source text and the "^" may wrap over multiple lines.

Previous: Eric SunshineNext: Junio C Hamano
Message 14 of 29 in “make chainlint output more newcomer-friendly”
  1. 0/2 make chainlint output more newcomer-friendlyEric Sunshine, Aug 29, 2024
  2. 1/2 chainlint: make error messages self-explanatoryEric Sunshine, Aug 29, 2024
  3. Patrick SteinhardtAug 29, 2024
  4. Jeff KingAug 29, 2024
  5. Eric SunshineAug 29, 2024
  6. Eric SunshineAug 29, 2024
  7. Junio C HamanoAug 29, 2024
  8. Eric SunshineAug 29, 2024
  9. Junio C HamanoAug 30, 2024
  10. 2/2 chainlint: reduce annotation noise-factorEric Sunshine, Aug 29, 2024
  11. Patrick SteinhardtAug 29, 2024
  12. Jeff KingAug 29, 2024
  13. Eric SunshineAug 29, 2024
  14. Eric SunshineAug 29, 2024
  15. Junio C HamanoAug 29, 2024
  16. Eric SunshineAug 30, 2024
  17. Junio C HamanoAug 30, 2024
  18. 0/3 make chainlint output more newcomer-friendlyEric Sunshine, Sep 10, 2024
  19. 1/3 chainlint: don't be fooled by "?!...?!" in test bodyEric Sunshine, Sep 10, 2024
  20. Junio C HamanoSep 10, 2024
  21. 2/3 chainlint: make error messages self-explanatoryEric Sunshine, Sep 10, 2024
  22. Patrick SteinhardtSep 10, 2024
  23. 3/3 chainlint: reduce annotation noise-factorEric Sunshine, Sep 10, 2024
  24. Patrick SteinhardtSep 10, 2024
  25. Eric SunshineSep 10, 2024
  26. Junio C HamanoSep 10, 2024
  27. Eric SunshineSep 10, 2024
  28. Jeff KingSep 10, 2024
  29. Junio C HamanoSep 10, 2024

Read the whole thread, see it on lore, or plain text.

$ cat FOOTERMessages come from the public archive at lore.kernel.org/git, fetched every hour. The front page is chosen and written each morning by an AI editor and can be wrong; the threads themselves are the record. About and API. For agents: an MCP server at https://gitlist.dev/mcp, and any thread, story or person page as Markdown by adding .md to its URL (or sending Accept: text/markdown). Details in /llms.txt.