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

Re: [PATCH 3/4] Add 'promisor-remote' capability to protocol v2

From
Christian Couder <christian.couder@gmail.com>
Date
Sep 10, 2024, 16:31 UTC
Message-ID
<CAP8UFD34FP2CCaeeYU5maUiNB5m5pD3_neyEYFtf8GQ4xCZmvA@mail.gmail.com>
In-Reply-To
<ZrDYIFolRlERFdUT@tanuki>
On Mon, Aug 5, 2024 at 8:01 PM Patrick Steinhardt <ps@pks.im> wrote:
>
> On Wed, Jul 31, 2024 at 03:40:13PM +0200, Christian Couder wrote:
Show 31 quoted lines
> > +The server may advertise some promisor remotes it is using, if it's OK
> > +for the server that a client uses them too. In this case <pr-infos>
> > +should be of the form:
> > +
> > +     pr-infos = pr-info | pr-infos ";" pr-info
> > +
> > +     pr-info = "name=" pr-name | "name=" pr-name "," "url=" pr-url
> > +
> > +where `pr-name` is the name of a promisor remote, and `pr-url` the
> > +urlencoded URL of that promisor remote.
> > +
> > +In this case a client wanting to use one or more promisor remotes the
> > +server advertised should reply with "promisor-remote=<pr-names>" where
> > +<pr-names> should be of the form:
> > +
> > +     pr-names = pr-name | pr-names ";" pr-name
> > +
> > +where `pr-name` is the name of a promisor remote the server
> > +advertised.
> > +
> > +If the server prefers a client not to use any promisor remote the
> > +server uses, or if the server doesn't use any promisor remote, it
> > +should only advertise "promisor-remote" without any value or "=" sign
> > +after it.
> > +
> > +In this case, or if the client doesn't want to use any promisor remote
> > +the server advertised, the client should reply only "promisor-remote"
> > +without any value or "=" sign after it.
>
> Why does the client have to advertise anything if they don't want to use
> any of the promisor remotes?

I have tried to explain it in a reply to Taylor, but as you, Junio and others seem to prefer the capability not to be advertised at all when not used, I have changed this in version 2.

Show 9 quoted lines
> > +The "promisor.advertise" and "promisor.acceptFromServer" configuration
> > +options can be used on the server and client side respectively to
> > +control what they advertise or accept respectively. See the
> > +documentation of these configuration options for more information.
>
> One thing I'm not totally clear on is the consequence of this
> capability. What is the expected consequence if the client accepts one
> of the promisor remotes? What is the consequence if the client accepts
> none?

I have tried to improve the documentation significatively, especially according to Junio's suggestion, in version 2.

> In the former case I'd expect that the server is free to omit objects,
> but that isn't made explicit anywhere, I think.
Junio also suggested making it explicit so I have done that in version 2.
> Also, is there any
> mechanism that tells the client exactly which objects have been omitted?

I don't think it's necessary. Agreeing on which promisor remote (name and URL) to use should be enough security wise. When using bundle-uri, for example, the server is not telling the client exactly which objects are in the bundle.

> In the latter case I assume that the result will be a full clone, that
> is the server fetched any objects it didn't have from the promisor?

Yeah, the server should fetch any objects it doesn't have from the promisor, so it can send everything to the client.

> Or does the server side continue to only honor whatever the client has
> provided as object filters, but signals to the client that it shall
> please contact somebody else when backfilling those promised objects?

No. Options to enable things like this could be built on top later if needed though.

Thanks for the review.
Previous: Junio C HamanoNext: Christian Couder
Message 25 of 110 in “Introduce a "promisor-remote" capability”
  1. 0/4 Introduce a "promisor-remote" capabilityChristian Couder, Jul 31, 2024
  2. 1/4 version: refactor strbuf_sanitize()Christian Couder, Jul 31, 2024
  3. Junio C HamanoJul 31, 2024
  4. Christian CouderAug 20, 2024
  5. 2/4 strbuf: refactor strbuf_trim_trailing_ch()Christian Couder, Jul 31, 2024
  6. Junio C HamanoJul 31, 2024
  7. Taylor BlauJul 31, 2024
  8. Christian CouderAug 20, 2024
  9. Christian CouderAug 20, 2024
  10. 3/4 Add 'promisor-remote' capability to protocol v2Christian Couder, Jul 31, 2024
  11. Taylor BlauJul 31, 2024
  12. Christian CouderAug 20, 2024
  13. Junio C HamanoAug 20, 2024
  14. Christian CouderSep 10, 2024
  15. Taylor BlauJul 31, 2024
  16. Christian CouderAug 20, 2024
  17. Junio C HamanoAug 20, 2024
  18. Christian CouderSep 10, 2024
  19. Junio C HamanoSep 10, 2024
  20. Junio C HamanoJul 31, 2024
  21. Junio C HamanoJul 31, 2024
  22. Christian CouderAug 20, 2024
  23. Patrick SteinhardtAug 5, 2024
  24. Junio C HamanoAug 19, 2024
  25. Christian CouderSep 10, 2024
  26. 4/4 promisor-remote: check advertised name or URLChristian Couder, Jul 31, 2024
  27. Junio C HamanoJul 31, 2024
  28. Christian CouderSep 10, 2024
  29. Junio C HamanoJul 31, 2024
  30. Taylor BlauJul 31, 2024
  31. 0/4 Introduce a "promisor-remote" capabilityChristian Couder, Sep 10, 2024
  32. 1/4 version: refactor strbuf_sanitize()Christian Couder, Sep 10, 2024
  33. 2/4 strbuf: refactor strbuf_trim_trailing_ch()Christian Couder, Sep 10, 2024
  34. 3/4 Add 'promisor-remote' capability to protocol v2Christian Couder, Sep 10, 2024
  35. Patrick SteinhardtSep 30, 2024
  36. Christian CouderSep 30, 2024
  37. Patrick SteinhardtOct 1, 2024
  38. Junio C HamanoOct 1, 2024
  39. Patrick SteinhardtNov 6, 2024
  40. Junio C HamanoNov 28, 2024
  41. Christian CouderNov 28, 2024
  42. Junio C HamanoNov 29, 2024
  43. 4/4 promisor-remote: check advertised name or URLChristian Couder, Sep 10, 2024
  44. Patrick SteinhardtSep 30, 2024
  45. Junio C HamanoSep 26, 2024
  46. Christian CouderSep 27, 2024
  47. Junio C HamanoSep 27, 2024
  48. rsbecker@nexbridge.comSep 27, 2024
  49. Kristoffer HaugsbakkSep 28, 2024
  50. Patrick SteinhardtSep 30, 2024
  51. Christian CouderSep 30, 2024
  52. Junio C HamanoSep 30, 2024
  53. Patrick SteinhardtOct 1, 2024
  54. Junio C HamanoSep 30, 2024
  55. brian m. carlsonSep 30, 2024
  56. Junio C HamanoSep 30, 2024
  57. Patrick SteinhardtOct 1, 2024
  58. 0/5 Introduce a "promisor-remote" capabilityChristian Couder, Dec 6, 2024
  59. 1/5 version: refactor strbuf_sanitize()Christian Couder, Dec 6, 2024
  60. Junio C HamanoDec 7, 2024
  61. Christian CouderJan 27, 2025
  62. 2/5 strbuf: refactor strbuf_trim_trailing_ch()Christian Couder, Dec 6, 2024
  63. Junio C HamanoDec 7, 2024
  64. Christian CouderJan 27, 2025
  65. karthik nayakDec 16, 2024
  66. 3/5 Add 'promisor-remote' capability to protocol v2Christian Couder, Dec 6, 2024
  67. Junio C HamanoDec 7, 2024
  68. Christian CouderJan 27, 2025
  69. 4/5 promisor-remote: check advertised name or URLChristian Couder, Dec 6, 2024
  70. 5/5 doc: add technical design doc for large object promisorsChristian Couder, Dec 6, 2024
  71. Junio C HamanoDec 10, 2024
  72. Christian CouderJan 27, 2025
  73. Junio C HamanoDec 10, 2024
  74. Patrick SteinhardtDec 16, 2024
  75. Christian CouderJan 27, 2025
  76. Junio C HamanoJan 27, 2025
  77. Christian CouderFeb 18, 2025
  78. Junio C HamanoDec 9, 2024
  79. Christian CouderDec 9, 2024
  80. Christian CouderDec 9, 2024
  81. Junio C HamanoDec 9, 2024
  82. Christian CouderJan 27, 2025
  83. Junio C HamanoJan 27, 2025
  84. 0/6 Introduce a "promisor-remote" capabilityChristian Couder, Jan 27, 2025
  85. 1/6 version: replace manual ASCII checks with isprint() for clarityChristian Couder, Jan 27, 2025
  86. 2/6 version: refactor redact_non_printables()Christian Couder, Jan 27, 2025
  87. 3/6 version: make redact_non_printables() non-staticChristian Couder, Jan 27, 2025
  88. Patrick SteinhardtJan 30, 2025
  89. Christian CouderFeb 18, 2025
  90. 4/6 Add 'promisor-remote' capability to protocol v2Christian Couder, Jan 27, 2025
  91. Patrick SteinhardtJan 30, 2025
  92. Christian CouderFeb 18, 2025
  93. 5/6 promisor-remote: check advertised name or URLChristian Couder, Jan 27, 2025
  94. Junio C HamanoJan 27, 2025
  95. Junio C HamanoJan 28, 2025
  96. Patrick SteinhardtJan 30, 2025
  97. Christian CouderFeb 18, 2025
  98. Christian CouderFeb 18, 2025
  99. 6/6 doc: add technical design doc for large object promisorsChristian Couder, Jan 27, 2025
  100. Junio C HamanoJan 27, 2025
  101. Christian CouderFeb 18, 2025
  102. 0/3 Introduce a "promisor-remote" capabilityChristian Couder, Feb 18, 2025
  103. 1/3 Add 'promisor-remote' capability to protocol v2Christian Couder, Feb 18, 2025
  104. 2/3 promisor-remote: check advertised name or URLChristian Couder, Feb 18, 2025
  105. 3/3 doc: add technical design doc for large object promisorsChristian Couder, Feb 18, 2025
  106. Patrick SteinhardtFeb 21, 2025
  107. Junio C HamanoMar 3, 2025
  108. Junio C HamanoFeb 18, 2025
  109. Patrick SteinhardtFeb 21, 2025
  110. Junio C HamanoFeb 21, 2025

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.