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

Re: [PATCH v3 06/10] clone: delay cloning until after remote HEAD checking

From
Junio C Hamano <gitster@pobox.com>
Date
Jan 24, 2012, 00:39 UTC
Message-ID
<7vvco253xx.fsf@alter.siamese.dyndns.org>
In-Reply-To
<7vzkde546x.fsf@alter.siamese.dyndns.org>
Junio C Hamano <gitster@pobox.com> writes:
Show 19 quoted lines
> Junio C Hamano <gitster@pobox.com> writes:
>
>> It breaks pushing to multiple URLs like this:
>>
>>     $ cat .git/config
>>     [remote "origin"]
>>             url = https://code.google.com/p/git-htmldocs/
>>             url = github.com:gitster/git-htmldocs.git
>>             push = refs/heads/master:refs/heads/master
>>     $ git push
>>
>> The second url that is supposed to use the git-over-ssh transport
>> mistakenly use https:// and fails with:
>>
>>     error: Couldn't resolve host 'github.com:gitster' while accessing
>>     github.com:gitster/git-htmldocs.git/info/refs
>>     fatal: HTTP request failed
>
> And here is an obvious band-aid to work it around.
-- >8 --
Subject: [PATCH] push: do not let configured foreign-vcs permanently clobbered

Recently, 6f48d39 (clone: delay cloning until after remote HEAD checking, 2012-01-16) tried to record if a remote helper needs to be called after parsing the remote when transport_get() is called, by overwriting the field meant to store the configured remote helper name in the remote structure.

This is OK when a remote represents a single remote repository, but fails miserably when pushing to locations with multiple URLs, like this:

    $ cat .git/config
    [remote "origin"]
        url = https://code.google.com/p/git-htmldocs/
        url = github.com:gitster/git-htmldocs.git
        push = refs/heads/master:refs/heads/master
    $ git push

The second url that is supposed to use the git-over-ssh transport mistakenly use https:// and fails with:

    error: Couldn't resolve host 'github.com:gitster' while accessing
    github.com:gitster/git-htmldocs.git/info/refs
    fatal: HTTP request failed

The right solution would probably be to dedicate a separate field to store the detected external helper to be used, which is valid only during a single use of transport until it is disconnected, instead of overwriting foreign_vcs field, but in the meantime, this band-aid should suffice.

Signed-off-by: Junio C Hamano <gitster@pobox.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
---
 builtin/push.c |    2 ++
 1 files changed, 2 insertions(+), 0 deletions(-)
diff --git a/builtin/push.c b/builtin/push.c
index 35cce53..5fb98a0 100644
--- a/builtin/push.c
+++ b/builtin/push.c
@@ -204,11 +204,13 @@ static int do_push(const char *repo, int flags)
 		url_nr = remote->url_nr;
 	}
 	if (url_nr) {
+		const char *configured_foreign_vcs = remote->foreign_vcs;
 		for (i = 0; i < url_nr; i++) {
 			struct transport *transport =
 				transport_get(remote, url[i]);
 			if (push_with_options(transport, flags))
 				errs++;
+			remote->foreign_vcs = configured_foreign_vcs;
 		}
 	} else {
 		struct transport *transport =
-- 
1.7.9.rc2.100.gfd863d
Previous: Junio C HamanoNext: Nguyen Thai Ngoc Duy
Message 60 of 68 in “clone: allow detached checkout when --branch takes a tag”
  1. clone: allow detached checkout when --branch takes a tagNguyễn Thái Ngọc Duy, Jan 5, 2012
  2. Jeff KingJan 5, 2012
  3. Nguyen Thai Ngoc DuyJan 6, 2012
  4. Jeff KingJan 6, 2012
  5. Junio C HamanoJan 5, 2012
  6. Nguyen Thai Ngoc DuyJan 6, 2012
  7. 1/6 t5601: add &&Nguyễn Thái Ngọc Duy, Jan 8, 2012
  8. 2/6 clone: write detached HEAD in bare repositoriesNguyễn Thái Ngọc Duy, Jan 8, 2012
  9. 3/6 clone: factor out checkout codeNguyễn Thái Ngọc Duy, Jan 8, 2012
  10. Junio C HamanoJan 10, 2012
  11. Nguyen Thai Ngoc DuyJan 10, 2012
  12. Junio C HamanoJan 10, 2012
  13. Nguyen Thai Ngoc DuyJan 10, 2012
  14. 4/6 clone: --branch=<branch> always means refs/heads/<branch>Nguyễn Thái Ngọc Duy, Jan 8, 2012
  15. Junio C HamanoJan 10, 2012
  16. 5/6 clone: allow --branch to take a tagNguyễn Thái Ngọc Duy, Jan 8, 2012
  17. 6/6 clone: print advice on checking out detached HEADNguyễn Thái Ngọc Duy, Jan 8, 2012
  18. Junio C HamanoJan 10, 2012
  19. Nguyen Thai Ngoc DuyJan 10, 2012
  20. Junio C HamanoJan 10, 2012
  21. Nguyen Thai Ngoc DuyJan 10, 2012
  22. 00/10 nd/clone-detachedNguyễn Thái Ngọc Duy, Jan 10, 2012
  23. 00/10 nd/clone-detachedNguyễn Thái Ngọc Duy, Jan 13, 2012
  24. Junio C HamanoJan 13, 2012
  25. Nguyen Thai Ngoc DuyJan 14, 2012
  26. Junio C HamanoJan 14, 2012
  27. Nguyen Thai Ngoc DuyJan 14, 2012
  28. Junio C HamanoJan 15, 2012
  29. 00/10 nd/clone-detached rebase onto nd/clone-single-branchNguyễn Thái Ngọc Duy, Jan 16, 2012
  30. 01/10 t5601: add missing && cascadeNguyễn Thái Ngọc Duy, Jan 16, 2012
  31. 02/10 clone: write detached HEAD in bare repositoriesNguyễn Thái Ngọc Duy, Jan 16, 2012
  32. 03/10 clone: factor out checkout codeNguyễn Thái Ngọc Duy, Jan 16, 2012
  33. 04/10 clone: factor out HEAD update codeNguyễn Thái Ngọc Duy, Jan 16, 2012
  34. 05/10 clone: factor out remote ref writingNguyễn Thái Ngọc Duy, Jan 16, 2012
  35. 06/10 clone: delay cloning until after remote HEAD checkingNguyễn Thái Ngọc Duy, Jan 16, 2012
  36. 07/10 clone: --branch=<branch> always means refs/heads/<branch>Nguyễn Thái Ngọc Duy, Jan 16, 2012
  37. 08/10 clone: refuse to clone if --branch points to bogus refNguyễn Thái Ngọc Duy, Jan 16, 2012
  38. 09/10 clone: allow --branch to take a tagNguyễn Thái Ngọc Duy, Jan 16, 2012
  39. 10/10 clone: print advice on checking out detached HEADNguyễn Thái Ngọc Duy, Jan 16, 2012
  40. 01/10 t5601: add missing && cascadeNguyễn Thái Ngọc Duy, Jan 13, 2012
  41. 02/10 clone: write detached HEAD in bare repositoriesNguyễn Thái Ngọc Duy, Jan 13, 2012
  42. 03/10 clone: factor out checkout codeNguyễn Thái Ngọc Duy, Jan 13, 2012
  43. 04/10 clone: factor out HEAD update codeNguyễn Thái Ngọc Duy, Jan 13, 2012
  44. 05/10 clone: factor out remote ref writingNguyễn Thái Ngọc Duy, Jan 13, 2012
  45. 06/10 clone: delay cloning until after remote HEAD checkingNguyễn Thái Ngọc Duy, Jan 13, 2012
  46. 07/10 clone: --branch=<branch> always means refs/heads/<branch>Nguyễn Thái Ngọc Duy, Jan 13, 2012
  47. 08/10 clone: refuse to clone if --branch points to bogus refNguyễn Thái Ngọc Duy, Jan 13, 2012
  48. 09/10 clone: allow --branch to take a tagNguyễn Thái Ngọc Duy, Jan 13, 2012
  49. 10/10 clone: print advice on checking out detached HEADNguyễn Thái Ngọc Duy, Jan 13, 2012
  50. 01/10 t5601: add missing && cascadeNguyễn Thái Ngọc Duy, Jan 10, 2012
  51. 02/10 clone: write detached HEAD in bare repositoriesNguyễn Thái Ngọc Duy, Jan 10, 2012
  52. 03/10 clone: factor out checkout codeNguyễn Thái Ngọc Duy, Jan 10, 2012
  53. 04/10 clone: factor out HEAD update codeNguyễn Thái Ngọc Duy, Jan 10, 2012
  54. 05/10 clone: factor out remote ref writingNguyễn Thái Ngọc Duy, Jan 10, 2012
  55. 06/10 clone: delay cloning until after remote HEAD checkingNguyễn Thái Ngọc Duy, Jan 10, 2012
  56. Junio C HamanoJan 11, 2012
  57. Nguyen Thai Ngoc DuyJan 12, 2012
  58. Junio C HamanoJan 24, 2012
  59. Junio C HamanoJan 24, 2012
  60. Junio C HamanoJan 24, 2012
  61. Nguyen Thai Ngoc DuyJan 24, 2012
  62. 07/10 clone: --branch=<branch> always means refs/heads/<branch>Nguyễn Thái Ngọc Duy, Jan 10, 2012
  63. Junio C HamanoJan 11, 2012
  64. Nguyen Thai Ngoc DuyJan 12, 2012
  65. 08/10 clone: refuse to clone if --branch points to bogus refNguyễn Thái Ngọc Duy, Jan 10, 2012
  66. 09/10 clone: allow --branch to take a tagNguyễn Thái Ngọc Duy, Jan 10, 2012
  67. Junio C HamanoJan 11, 2012
  68. 10/10 clone: print advice on checking out detached HEADNguyễn Thái Ngọc Duy, Jan 10, 2012

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.