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

Re: [PATCH 1/2] mingw: verify that paths are not mistaken for remote nicknames

From
Ævar Arnfjörð Bjarmason <avarab@gmail.com>
Date
May 29, 2017, 20:40 UTC
Message-ID
<CACBZZX5BgVbUFEwQjHqJVfHFUitTJRaCx-Lg+j7q54uErEpqcg@mail.gmail.com>
In-Reply-To
<17b3d445-0cc7-2000-dc67-11fb53c7040b@kdbg.org>
On Mon, May 29, 2017 at 10:25 PM, Johannes Sixt <j6t@kdbg.org> wrote:
Show 78 quoted lines
> Am 26.05.2017 um 05:35 schrieb Junio C Hamano:
>> When asked to open/fopen a path, e.g. "a/b:/c", which does not exist
>> on the filesystem, Windows (correctly) fails to open it but sets
>> EINVAL to errno because the pathname has characters that cannot be
>> stored in its filesystem.
>>
>> As this is an expected failure, teach is_missing_file_error() helper
>> about this case.
>>
>> This is RFC, as there may be a case where we get EINVAL from
>> open/fopen for reasons other than "the filesystem does not like this
>> pathname" that may be worth reporting to the user, and this change
>> is sweeping such an error under the rug.
>>
>> Signed-off-by: Junio C Hamano <gitster@pobox.com>
>> ---
>>   wrapper.c | 4 ++++
>>   1 file changed, 4 insertions(+)
>>
>> diff --git a/wrapper.c b/wrapper.c
>> index f1c87ec7ea..74aa3b7803 100644
>> --- a/wrapper.c
>> +++ b/wrapper.c
>> @@ -434,6 +434,10 @@ static void warn_on_inaccessible(const char *path)
>>    * see if the errno indicates a missing file that we can safely ignore.
>>    */
>>   static int is_missing_file_error(int errno_) {
>> +#ifdef GIT_WINDOWS_NATIVE
>> +     if (errno_ == EINVAL)
>> +             return 1;
>> +#endif
>>       return (errno_ == ENOENT || errno_ == ENOTDIR);
>>   }
>
> I would prefer to catch the case in the compatibility layer. Here is
> a two patch series that would replace your 12/13 and 13/13.
>
> ---- 8< ----
> From: Johannes Schindelin <Johannes.Schindelin@gmx.de>
> Subject: mingw: verify that paths are not mistaken for remote nicknames
>
> This added test case simply verifies that users will not be bothered
> with bogus complaints à la
>
>         warning: unable to access '.git/remotes/D:\repo': Invalid argument
>
> when fetching from a Windows path (in this case, D:\repo).
>
> [j6t: mark the new test as test_expect_failure]
>
> Signed-off-by: Johannes Schindelin <johannes.schindelin@gmx.de>
> Signed-off-by: Johannes Sixt <j6t@kdbg.org>
> ---
>  t/t5580-clone-push-unc.sh | 10 ++++++++--
>  1 file changed, 8 insertions(+), 2 deletions(-)
>
> diff --git a/t/t5580-clone-push-unc.sh b/t/t5580-clone-push-unc.sh
> index b195f71ea9..fd719a209e 100755
> --- a/t/t5580-clone-push-unc.sh
> +++ b/t/t5580-clone-push-unc.sh
> @@ -1,13 +1,19 @@
>  #!/bin/sh
>
> -test_description='various UNC path tests (Windows-only)'
> +test_description='various Windows-only path tests'
>  . ./test-lib.sh
>
>  if ! test_have_prereq MINGW; then
> -       skip_all='skipping UNC path tests, requires Windows'
> +       skip_all='skipping Windows-only path tests'
>         test_done
>  fi
>
> +test_expect_failure 'remote nick cannot contain backslashes' '
> +       BACKSLASHED="$(pwd | tr / \\\\)" &&
> +       git ls-remote "$BACKSLASHED" >out 2>err &&
> +       ! grep "unable to access" err
> +'
Doesn't this need test_i18ngrep?:
    wrapper.c:581:  warning_errno(_("unable to access '%s'"), path);
    wrapper.c:602:          die_errno(_("unable to access '%s'"), path);
Or is it one of these, which isn't translated:
    http-push.c:1531:               error("unable to access '%s': %s",
url, curl_errorstr);
    remote-curl.c:319:              die("unable to access '%s': %s",
url.buf, curl_errorstr);
Show 6 quoted lines
> +
>  UNCPATH="$(pwd)"
>  case "$UNCPATH" in
>  [A-Z]:*)
> --
> 2.13.0.55.g17b7d13330
Previous: Johannes SixtNext: Johannes Sixt
Message 30 of 41 in “What's cooking in git.git (May 2017, #06; Mon, 22)”
  1. Junio C HamanoMay 22, 2017
  2. Jonathan NiederMay 22, 2017
  3. Junio C HamanoMay 23, 2017
  4. Jonathan NiederMay 23, 2017
  5. Junio C HamanoMay 23, 2017
  6. Duy NguyenMay 25, 2017
  7. Junio C HamanoMay 26, 2017
  8. 00/13 reporting unexpected errors after (f)openJunio C Hamano, May 26, 2017
  9. 01/13 git_fopen: fix a sparse 'not declared' warningJunio C Hamano, May 26, 2017
  10. 02/13 use xfopen() in more placesJunio C Hamano, May 26, 2017
  11. 03/13 clone: use xfopen() instead of fopen()Junio C Hamano, May 26, 2017
  12. 05/13 config.mak.uname: set FREAD_READS_DIRECTORIES for Darwin, tooJunio C Hamano, May 26, 2017
  13. Stefan BellerMay 30, 2017
  14. Junio C HamanoMay 30, 2017
  15. Stefan BellerMay 30, 2017
  16. 04/13 config.mak.uname: set FREAD_READS_DIRECTORIES for Linux and FreeBSDJunio C Hamano, May 26, 2017
  17. 06/13 wrapper.c: add and use warn_on_fopen_errors()Junio C Hamano, May 26, 2017
  18. 07/13 wrapper.c: add and use fopen_or_warn()Junio C Hamano, May 26, 2017
  19. 09/13 print errno when reporting a system call errorJunio C Hamano, May 26, 2017
  20. 10/13 rerere.c: move error_errno() closer to the source system callJunio C Hamano, May 26, 2017
  21. Stefan BellerMay 30, 2017
  22. 08/13 wrapper.c: make warn_on_inaccessible() staticJunio C Hamano, May 26, 2017
  23. 11/13 log: fix memory leak in open_next_file()Junio C Hamano, May 26, 2017
  24. 12/13 wrapper: factor out is_missing_file_error()Junio C Hamano, May 26, 2017
  25. 1/2 compat-util: is_missing_file_error()Junio C Hamano, May 30, 2017
  26. 2/2 treewide: use is_missing_file_error() where ENOENT and ENOTDIR are checkedJunio C Hamano, May 30, 2017
  27. 13/13 is_missing_file_error(): work around EINVAL on WindowsJunio C Hamano, May 26, 2017
  28. 1/2 mingw: verify that paths are not mistaken for remote nicknamesJohannes Sixt, May 29, 2017
  29. 2/2 mingw_fopen: report ENOENT for invalid file namesJohannes Sixt, May 29, 2017
  30. Ævar Arnfjörð BjarmasonMay 29, 2017
  31. Johannes SixtMay 29, 2017
  32. Ramsay JonesMay 29, 2017
  33. Junio C HamanoMay 30, 2017
  34. Ramsay JonesMay 30, 2017
  35. Junio C HamanoMay 29, 2017
  36. Junio C HamanoMay 30, 2017
  37. Johannes SixtMay 30, 2017
  38. Junio C HamanoMay 30, 2017
  39. Stefan BellerMay 30, 2017
  40. Junio C HamanoMay 30, 2017
  41. Stefan BellerMay 30, 2017

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.