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

Re: [PATCH v3 1/9] t5520: fixup file contents comparisons

From
Junio C Hamano <gitster@pobox.com>
Date
May 15, 2015, 18:37 UTC
Message-ID
<xmqq7fs9hekc.fsf@gitster.dls.corp.google.com>
In-Reply-To
<CACRoPnSbekLANNiGOyxN70TCUd1c=wcrU_6Gfew5pp5EBpSEsA@mail.gmail.com>
Paul Tan <pyokagan@gmail.com> writes:
Show 21 quoted lines
> Hi Junio,
>
> On Fri, May 15, 2015 at 1:44 AM, Junio C Hamano <gitster@pobox.com> wrote:
>> Change that 'verbose test' line to
>>
>>         verbose test 1 = $(find .git/rebase-apply -name "000*" | wc -l)
>>
>> i.e. losing the double-quotes around $().
>
> Noted and fixed. Interesting quirk though :-).
>
>> By the way, thanks for a fine demonstration that the 'verbose test'
>> is not very useful.
>>
>> This output
>>
>>> command failed:  'test' '1' '=' '       1'
>
> Personally, I find that the quoting provided by "verbose" helps make
> it clear that it's a whitespace issue, which might be a bit harder to
> spot with the output of set -x I think.

To be fair, yes, because of the leading SP in the RHS, I immediately knew that this was a "wc -l" from that without running the test one more time with "-i -v -x". The "rev-parse --sq-quote" did help. Without the --sq-quote trick, i.e. "command failed: test 1 = 1", would actually make the debugger suspect that there would be a quoting issue anyway, so it is not a very big deal, though.

In any case, that "test 1 = 1" (with or without quoting) helped only because I had to deal with "wc -l" issues in the past. Without telling how that ' 1' ended up compared with '1' by showing "wc -l" on that 'command failed:' line, it wouldn't have helped much if the debugger were not me.

> Other than that, I'm also convinced that "verbose" doesn't really
> offer much. Will remove in the re-roll.

Just to avoid misunderstanding, please do not remove 'verbose ' blindly without thinking while doing so, as you already did 1/3 of the necessary job to make things better.

You might have noticed, while adding them, there were something common that we currently do with a bare 'test' only because we haven't identified common needs. As I already said, it may be that we often try to see a file has a known single line content (I didn't check if that were the case; I am just giving you an example) and only because there is no ready-made test_file_contents helper to be used, the current tests say

	test expected_string = "$(cat file)"

And if that were the case, it is a good thing to have a new helper like this

	test_file_contents () {
		if test "$(cat "$1")" != "$2"
		then
			echo "Contents of file '$1' is not '$2'"
                        false
		fi
	}
in t/test-lib-functions.sh and convert them to say
	test_file_contents file expected_string
That would be an improvement (and that is the remaining 2/3 ;-).
Thanks.
Previous: Paul TanNext: Junio C Hamano
Message 8 of 25 in “Improve git-pull test coverage”
  1. 0/9 Improve git-pull test coveragePaul Tan, May 13, 2015
  2. 1/9 t5520: fixup file contents comparisonsPaul Tan, May 13, 2015
  3. Junio C HamanoMay 13, 2015
  4. Junio C HamanoMay 13, 2015
  5. Michael BlumeMay 14, 2015
  6. Junio C HamanoMay 14, 2015
  7. Paul TanMay 15, 2015
  8. Junio C HamanoMay 15, 2015
  9. Junio C HamanoMay 15, 2015
  10. Paul TanMay 16, 2015
  11. Junio C HamanoMay 16, 2015
  12. Junio C HamanoMay 16, 2015
  13. Paul TanMay 17, 2015
  14. 2/9 t5520: ensure origin refs are updatedPaul Tan, May 13, 2015
  15. Junio C HamanoMay 13, 2015
  16. Paul TanMay 18, 2015
  17. 3/9 t5520: test no merge candidates casesPaul Tan, May 13, 2015
  18. 4/9 t5520: test for failure if index has unresolved entriesPaul Tan, May 13, 2015
  19. Matthieu MoyMay 13, 2015
  20. Paul TanMay 15, 2015
  21. 5/9 t5520: test work tree fast-forward when fetch updates headPaul Tan, May 13, 2015
  22. 6/9 t5520: test --rebase with multiple branchesPaul Tan, May 13, 2015
  23. 7/9 t5520: test --rebase failure on unborn branch with indexPaul Tan, May 13, 2015
  24. 8/9 t5521: test --dry-run does not make any changesPaul Tan, May 13, 2015
  25. 9/9 t5520: check reflog action in fast-forward mergePaul Tan, May 13, 2015

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.