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

Re: [PATCH 8/9] update-ref: read commands in a line-wise fashion

From
Junio C Hamano <gitster@pobox.com>
Date
Mar 30, 2020, 17:39 UTC
Message-ID
<xmqqk13191yu.fsf@gitster.c.googlers.com>
In-Reply-To
<20200330081144.GB186019@ncase.pks.im>
Patrick Steinhardt <ps@pks.im> writes:
Show 23 quoted lines
> On Fri, Mar 27, 2020 at 02:58:38PM -0700, Junio C Hamano wrote:
>> Patrick Steinhardt <ps@pks.im> writes:
>> 
>> >  static const struct parse_cmd {
>> >  	const char *prefix;
>> > -	const char *(*fn)(struct ref_transaction *, const char *, const char *);
>> > +	void (*fn)(struct ref_transaction *, const char *, const char *);
>> > +	unsigned extra_lines;
>> 
>> Define and explain the meaning of extra_lines in a comment.  Without
>> it, the most natural way to infer what the variable means by readers
>> would be that "update" command sometimes receives up to two more
>> lines but it also is perfectly normal if there is no extra line.
>> 
>> But I am not sure if that is what is going on.
>> 
>> "update" _always_ needs exactly two more input "lines" when the
>> input is NUL delimited, perhaps, and it is an _error_ if there are
>> not these two lines, no?
>
> That's extactly right. I pondered on a good name, but I wasn't yet able
> to come up with any one that brings across its meaning. I originally had
> `extra_args` but changed it after some internal discussion with Chris.

Wouldn't it make more sense to store the number of args expected here, not "extra"? The one that does not take any then can say "I am subcommand X and I do not want a SP after my name".

The "sometimes after the command there is line.termination, and sometimes there is NUL, and yet some other times there is SP", which is a sloppy way to check malformed input (i.e. you do not want SP for a command that does not take parameters), that you have after the loop can then go, if you did the parsing that way.

Previous: Patrick SteinhardtNext: Patrick Steinhardt
Message 21 of 48 in “Support for transactions in `git-update-ref --stdin`”
  1. 0/9 Support for transactions in `git-update-ref --stdin`Patrick Steinhardt, Mar 25, 2020
  2. 1/9 refs: fix segfault when aborting empty transactionPatrick Steinhardt, Mar 25, 2020
  3. Junio C HamanoMar 27, 2020
  4. 2/9 git-update-ref.txt: add missing wordPatrick Steinhardt, Mar 25, 2020
  5. 3/9 strbuf: provide function to append whole linesPatrick Steinhardt, Mar 25, 2020
  6. Junio C HamanoMar 27, 2020
  7. Patrick SteinhardtMar 30, 2020
  8. Junio C HamanoMar 30, 2020
  9. 4/9 update-ref: organize commands in an arrayPatrick Steinhardt, Mar 25, 2020
  10. Junio C HamanoMar 27, 2020
  11. Patrick SteinhardtMar 30, 2020
  12. Junio C HamanoMar 30, 2020
  13. Patrick SteinhardtMar 30, 2020
  14. 5/9 update-ref: drop unused argument for `parse_refname`Patrick Steinhardt, Mar 25, 2020
  15. 6/9 update-ref: pass end pointer instead of strbufPatrick Steinhardt, Mar 25, 2020
  16. 7/9 update-ref: move transaction handling into `update_refs_stdin()`Patrick Steinhardt, Mar 25, 2020
  17. Junio C HamanoMar 27, 2020
  18. 8/9 update-ref: read commands in a line-wise fashionPatrick Steinhardt, Mar 25, 2020
  19. Junio C HamanoMar 27, 2020
  20. Patrick SteinhardtMar 30, 2020
  21. Junio C HamanoMar 30, 2020
  22. 9/9 update-ref: implement interactive transaction handlingPatrick Steinhardt, Mar 25, 2020
  23. Junio C HamanoMar 27, 2020
  24. 0/9 Support for transactions in `git-update-ref --stdin`Patrick Steinhardt, Mar 30, 2020
  25. 1/9 refs: fix segfault when aborting empty transactionPatrick Steinhardt, Mar 30, 2020
  26. 2/9 git-update-ref.txt: add missing wordPatrick Steinhardt, Mar 30, 2020
  27. 3/9 strbuf: provide function to append whole linesPatrick Steinhardt, Mar 30, 2020
  28. 4/9 update-ref: organize commands in an arrayPatrick Steinhardt, Mar 30, 2020
  29. 5/9 update-ref: drop unused argument for `parse_refname`Patrick Steinhardt, Mar 30, 2020
  30. 6/9 update-ref: pass end pointer instead of strbufPatrick Steinhardt, Mar 30, 2020
  31. 7/9 update-ref: move transaction handling into `update_refs_stdin()`Patrick Steinhardt, Mar 30, 2020
  32. 8/9 update-ref: read commands in a line-wise fashionPatrick Steinhardt, Mar 30, 2020
  33. 9/9 update-ref: implement interactive transaction handlingPatrick Steinhardt, Mar 30, 2020
  34. 0/9 Support for transactions in `git-update-ref --stdin`Patrick Steinhardt, Apr 2, 2020
  35. 1/9 refs: fix segfault when aborting empty transactionPatrick Steinhardt, Apr 2, 2020
  36. 2/9 git-update-ref.txt: add missing wordPatrick Steinhardt, Apr 2, 2020
  37. 3/9 strbuf: provide function to append whole linesPatrick Steinhardt, Apr 2, 2020
  38. 4/9 update-ref: organize commands in an arrayPatrick Steinhardt, Apr 2, 2020
  39. 5/9 update-ref: drop unused argument for `parse_refname`Patrick Steinhardt, Apr 2, 2020
  40. 6/9 update-ref: pass end pointer instead of strbufPatrick Steinhardt, Apr 2, 2020
  41. 7/9 update-ref: move transaction handling into `update_refs_stdin()`Patrick Steinhardt, Apr 2, 2020
  42. 8/9 update-ref: read commands in a line-wise fashionPatrick Steinhardt, Apr 2, 2020
  43. 9/9 update-ref: implement interactive transaction handlingPatrick Steinhardt, Apr 2, 2020
  44. Phillip WoodApr 3, 2020
  45. Patrick SteinhardtApr 3, 2020
  46. Junio C HamanoApr 3, 2020
  47. Junio C HamanoApr 3, 2020
  48. Patrick SteinhardtApr 6, 2020

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.