From: Jeff King <peff@peff.net>
To: "Ævar Arnfjörð Bjarmason" <avarab@gmail.com>
Cc: "Duy Nguyen" <pclouds@gmail.com>,
"Rohit Ashiwal via GitGitGadget" <gitgitgadget@gmail.com>,
"Git Mailing List" <git@vger.kernel.org>,
"Junio C Hamano" <gitster@pobox.com>,
"Rohit Ashiwal" <rohit.ashiwal265@gmail.com>,
"SZEDER Gábor" <szeder.dev@gmail.com>,
"Matthieu Moy" <git@matthieu-moy.fr>
Subject: Re: Do test-path_is_{file,dir,exists} make sense anymore with -x?
Date: Tue, 26 Feb 2019 12:35:43 -0500 [thread overview]
Message-ID: <20190226173542.GC19606@sigill.intra.peff.net> (raw)
In-Reply-To: <87sgwav8cp.fsf@evledraar.gmail.com>
On Tue, Feb 26, 2019 at 05:10:30PM +0100, Ævar Arnfjörð Bjarmason wrote:
> But 4 years after this was added in a136f6d8ff ("test-lib.sh: support -x
> option for shell-tracing", 2014-10-10) we got -x, and then with "-i -v -x":
>
> expecting success:
> test_path_is_dir .git &&
> test_path_is_file doesnotexist &&
> test_path_is_file .git/config
>
> + test_path_is_dir .git
> + test -d .git
> + test_path_is_file doesnotexist
> + test -f doesnotexist
> + echo File doesnotexist doesn't exist.
> File doesnotexist doesn't exist.
> + false
> error: last command exited with $?=1
> not ok 1 - check files
>
> But by just using "test -d/-e": the much shorter:
>
> + test -d .git
> + test -f doesnotexist
> error: last command exited with $?=1
> not ok 1 - check files
>
> So I wonder if these days we shouldn't do this the other way around and
> get rid of these. Every test_* wrapper we add adds a bit of cognitive
> overload when you have to remember Git's specific shellscript dialect.
I don't have a strong opinion, but I do agree that with "-x" it's nicer
without the wrappers. I typically re-run with just "-v" on a failure,
and only turn to "-x" if the verbose output isn't helpful. However, with
the rise of multi-platform CI jobs which try to collect as much
information as possible in the initial run, I do find myself looking at
"-x" more often.
As Gábor notes, you can't run every script with "-x". But I find it's
pretty consistent these days (and totally so if you have a recent bash).
I dunno. Maybe people on other platforms (who might not have bash) would
care more.
I had a vague notion that there was some reason (portability?) that we
preferred to have the wrappers. But as your patch shows, they really are
just calling "test" and nothing else.
> test_must_be_empty () {
> + # We don't want to remove this as noted in ec10b018e7 ("tests:
> + # use 'test_must_be_empty' instead of '! test -s'",
> + # 2018-08-19)
> test_path_is_file "$1" &&
> if test -s "$1"
> then
You'd still want it to become "test -f" though, right?
-Peff
next prev parent reply other threads:[~2019-02-26 17:35 UTC|newest]
Thread overview: 42+ messages / expand[flat|nested] mbox.gz Atom feed top
2019-02-26 13:42 [PATCH 0/1] [GSoC][PATCH] tests: replace test -(d|f) with test_path_is_(dir|file) Rohit Ashiwal via GitGitGadget
2019-02-26 13:42 ` [PATCH 1/1] tests: replace `test -(d|f)` " Rohit Ashiwal via GitGitGadget
2019-02-26 14:04 ` Duy Nguyen
2019-02-26 16:10 ` Do test-path_is_{file,dir,exists} make sense anymore with -x? Ævar Arnfjörð Bjarmason
2019-02-26 17:04 ` SZEDER Gábor
2019-02-26 17:43 ` Jeff King
2019-02-26 19:39 ` SZEDER Gábor
2019-02-26 21:01 ` Jeff King
2019-03-03 16:04 ` SZEDER Gábor
2019-03-05 4:55 ` Jeff King
2019-03-04 14:36 ` SZEDER Gábor
2019-03-05 4:58 ` Jeff King
2019-02-26 17:48 ` Matthieu Moy
2019-02-26 18:24 ` Jeff King
2019-02-26 17:35 ` Jeff King [this message]
2019-02-26 19:58 ` Johannes Schindelin
2019-02-26 21:02 ` Jeff King
2019-02-27 10:01 ` Duy Nguyen
2019-03-01 2:52 ` Junio C Hamano
2019-02-26 16:01 ` [PATCH 1/1] tests: replace `test -(d|f)` with test_path_is_(dir|file) Johannes Schindelin
2019-02-26 16:30 ` Martin Ågren
2019-02-26 18:29 ` Rohit Ashiwal
2019-02-26 19:52 ` Johannes Schindelin
2019-02-26 20:01 ` Rohit Ashiwal
2019-02-27 5:49 ` Martin Ågren
2019-02-26 14:26 ` [PATCH v2 0/1] [GSoC][PATCH] t3600: use test_path_is_dir and test_path_is_file Rohit Ashiwal via GitGitGadget
2019-02-26 14:26 ` [PATCH v2 1/1] " Rohit Ashiwal via GitGitGadget
2019-02-26 16:37 ` SZEDER Gábor
2019-02-26 18:40 ` Rohit Ashiwal
2019-02-26 20:02 ` Johannes Schindelin
2019-02-26 20:05 ` Rohit Ashiwal
2019-02-26 22:48 ` [PATCH v3 0/1] [GSoC][PATCH] t3600: use test_path_is_* helper functions Rohit Ashiwal via GitGitGadget
2019-02-26 22:48 ` [PATCH v3 1/1] t3600: use test_path_is_* functions Rohit Ashiwal via GitGitGadget
2019-02-27 10:12 ` Duy Nguyen
2019-02-28 10:26 ` [PATCH v4 0/1] [GSoC][PATCH] t3600: use test_path_is_* helper functions Rohit Ashiwal via GitGitGadget
2019-02-28 10:26 ` [PATCH v4 1/1] t3600: use test_path_is_* functions Rohit Ashiwal via GitGitGadget
2019-02-28 19:02 ` [GSoC] acknowledging mistakes Rohit Ashiwal
2019-03-01 2:51 ` Junio C Hamano
2019-03-01 13:13 ` Feeling confused a little bit Rohit Ashiwal
2019-03-02 4:24 ` Rafael Ascensão
2019-03-02 14:46 ` Thomas Gummerer
2019-03-02 16:21 ` [GSoC] Thanking Rohit Ashiwal
Reply instructions:
You may reply publicly to this message via plain-text email
using any one of the following methods:
* Save the following mbox file, import it into your mail client,
and reply-to-all from there: mbox
Avoid top-posting and favor interleaved quoting:
https://en.wikipedia.org/wiki/Posting_style#Interleaved_style
* Reply using the --to, --cc, and --in-reply-to
switches of git-send-email(1):
git send-email \
--in-reply-to=20190226173542.GC19606@sigill.intra.peff.net \
--to=peff@peff.net \
--cc=avarab@gmail.com \
--cc=git@matthieu-moy.fr \
--cc=git@vger.kernel.org \
--cc=gitgitgadget@gmail.com \
--cc=gitster@pobox.com \
--cc=pclouds@gmail.com \
--cc=rohit.ashiwal265@gmail.com \
--cc=szeder.dev@gmail.com \
/path/to/YOUR_REPLY
https://kernel.org/pub/software/scm/git/docs/git-send-email.html
* If your mail client supports setting the In-Reply-To header
via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line
before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).