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

Re: [PATCH 05/38] refs: create a base class "ref_store" for files_ref_store

From
Eric Sunshine <sunshine@sunshineco.com>
Date
Jun 10, 2016, 08:08 UTC
Message-ID
<CAPig+cQqnDFQ1=ydQReJimVhL7WrjH9CD1RZ63ddCnNm=YVenA@mail.gmail.com>
In-Reply-To
<8ee7d7359f6763cba29dc788aba2f236204bb76e.1464983301.git.mhagger@alum.mit.edu>
On Fri, Jun 3, 2016 at 5:03 PM, Michael Haggerty <mhagger@alum.mit.edu> wrote:
Show 31 quoted lines
> We want ref_stores to be polymorphic, so invent a base class of which
> files_ref_store is a derived class. For now there is a one-to-one
> relationship between ref_stores and submodules.
>
> Signed-off-by: Michael Haggerty <mhagger@alum.mit.edu>
> ---
> diff --git a/refs/files-backend.c b/refs/files-backend.c
> @@ -973,53 +967,54 @@ static void clear_loose_ref_cache(struct files_ref_store *refs)
> +/*
> + * Downcast ref_store to files_ref_store. Die if ref_store is not a
> + * files_ref_store. If submodule_allowed is not true, then also die if
> + * files_ref_store is for a submodule (i.e., not for the main
> + * repository). caller is used in any necessary error messages.
> + */
> +static struct files_ref_store *files_downcast(
> +               struct ref_store *ref_store, int submodule_allowed,
> +               const char *caller)
>  {
>         struct files_ref_store *refs;
>
> +       if (ref_store->be != &refs_be_files)
> +               die("BUG: ref_store is type \"%s\" not \"files\" in %s",
> +                   ref_store->be->name, caller);
>
> +       refs = (struct files_ref_store *)ref_store;
> +
> +       if (!submodule_allowed)
> +               assert_main_repository(ref_store, caller);
> +
> +       return refs;
>  }

Aside from returning the downcasted value, 'refs' doesn't seem to be used for anything, thus could be dropped and the downcasted value returned directly:

    return (struct files_ref_store *)ref_store;
Not worth a re-roll.
Show 17 quoted lines
> diff --git a/refs/refs-internal.h b/refs/refs-internal.h
> @@ -521,11 +521,89 @@ int read_raw_ref(const char *refname, unsigned char *sha1,
> +/*
> + * A representation of the reference store for the main repository or
> + * a submodule. The ref_store instances for submodules are kept in a
> + * linked list.
> + */
> +struct ref_store {
> +       /* The backend describing this ref_store's storage scheme: */
> +       const struct ref_storage_be *be;
> +
> +       /*
> +        * The name of the submodule represented by this object, or
> +        * the empty string if it represents the main repository's
> +        * reference store:
> +        */
> +       const char *submodule;
Tangent: Apart from backward compatibility due to all the existing
code which tests *submodule to distinguish between the main repository
and a submodule, is there a technical reason that this ought to store
an empty string rather than (the more idiomatic) NULL to signify the
main repository?
Previous: Junio C HamanoNext: Michael Haggerty
Message 17 of 67 in “Virtualization of the refs API”
  1. 00/38 Virtualization of the refs APIMichael Haggerty, Jun 3, 2016
  2. 01/38 resolve_gitlink_ref(): eliminate temporary variableMichael Haggerty, Jun 3, 2016
  3. 02/38 rename_ref_available(): add docstringMichael Haggerty, Jun 3, 2016
  4. Junio C HamanoJun 7, 2016
  5. Michael HaggertyJun 9, 2016
  6. Junio C HamanoJun 9, 2016
  7. 03/38 refs: rename struct ref_cache to files_ref_storeMichael Haggerty, Jun 3, 2016
  8. 04/38 refs: add a backend method structureMichael Haggerty, Jun 3, 2016
  9. 05/38 refs: create a base class "ref_store" for files_ref_storeMichael Haggerty, Jun 3, 2016
  10. Junio C HamanoJun 7, 2016
  11. René ScharfeJun 9, 2016
  12. Junio C HamanoJun 7, 2016
  13. Michael HaggertyJun 9, 2016
  14. Junio C HamanoJun 9, 2016
  15. Michael HaggertyJun 10, 2016
  16. Junio C HamanoJun 10, 2016
  17. Eric SunshineJun 10, 2016
  18. Michael HaggertyJun 10, 2016
  19. 06/38 add_packed_ref(): add a files_ref_store argumentMichael Haggerty, Jun 3, 2016
  20. 07/38 get_packed_ref(): add a files_ref_store argumentMichael Haggerty, Jun 3, 2016
  21. 08/38 resolve_missing_loose_ref(): add a files_ref_store argumentMichael Haggerty, Jun 3, 2016
  22. 09/38 {lock,commit,rollback}_packed_refs(): add files_ref_store argumentsMichael Haggerty, Jun 3, 2016
  23. 10/38 refs: add a transaction_commit() methodMichael Haggerty, Jun 3, 2016
  24. 11/38 refs: reorder definitionsMichael Haggerty, Jun 3, 2016
  25. 12/38 resolve_packed_ref(): rename function from resolve_missing_loose_ref()Michael Haggerty, Jun 3, 2016
  26. 13/38 resolve_gitlink_packed_ref(): remove functionMichael Haggerty, Jun 3, 2016
  27. 14/38 read_raw_ref(): take a (struct ref_store *) argumentMichael Haggerty, Jun 3, 2016
  28. 15/38 resolve_ref_recursively(): new functionMichael Haggerty, Jun 3, 2016
  29. 16/38 resolve_gitlink_ref(): implement using resolve_ref_recursively()Michael Haggerty, Jun 3, 2016
  30. Junio C HamanoJun 7, 2016
  31. Eric SunshineJun 14, 2016
  32. Michael HaggertyJun 16, 2016
  33. 17/38 resolve_gitlink_ref(): avoid memory allocation in many casesMichael Haggerty, Jun 3, 2016
  34. Junio C HamanoJun 7, 2016
  35. Michael HaggertyJun 9, 2016
  36. 18/38 resolve_gitlink_ref(): rename path parameter to submoduleMichael Haggerty, Jun 3, 2016
  37. 19/38 refs: make read_raw_ref() virtualMichael Haggerty, Jun 3, 2016
  38. 20/38 refs: make verify_refname_available() virtualMichael Haggerty, Jun 3, 2016
  39. 21/38 refs: make pack_refs() virtualMichael Haggerty, Jun 3, 2016
  40. Junio C HamanoJun 7, 2016
  41. Michael HaggertyJun 9, 2016
  42. 22/38 refs: make create_symref() virtualMichael Haggerty, Jun 3, 2016
  43. 23/38 refs: make peel_ref() virtualMichael Haggerty, Jun 3, 2016
  44. Junio C HamanoJun 7, 2016
  45. Michael HaggertyJun 9, 2016
  46. 24/38 repack_without_refs(): add a files_ref_store argumentMichael Haggerty, Jun 3, 2016
  47. 25/38 lock_raw_ref(): add a files_ref_store argumentMichael Haggerty, Jun 3, 2016
  48. 26/38 commit_ref_update(): add a files_ref_store argumentMichael Haggerty, Jun 3, 2016
  49. 27/38 lock_ref_for_update(): add a files_ref_store argumentMichael Haggerty, Jun 3, 2016
  50. 28/38 lock_ref_sha1_basic(): add a files_ref_store argumentMichael Haggerty, Jun 3, 2016
  51. 29/38 split_symref_update(): add a files_ref_store argumentMichael Haggerty, Jun 3, 2016
  52. 30/38 files_ref_iterator_begin(): take a ref_store argumentMichael Haggerty, Jun 3, 2016
  53. 31/38 refs: add method iterator_beginMichael Haggerty, Jun 3, 2016
  54. 32/38 refs: add methods for reflogMichael Haggerty, Jun 3, 2016
  55. 33/38 refs: add method for initial ref transaction commitMichael Haggerty, Jun 3, 2016
  56. 34/38 refs: add method for delete_refsMichael Haggerty, Jun 3, 2016
  57. Junio C HamanoJun 7, 2016
  58. Michael HaggertyJun 9, 2016
  59. 35/38 refs: add methods to init refs dbMichael Haggerty, Jun 3, 2016
  60. 36/38 refs: add method to rename refsMichael Haggerty, Jun 3, 2016
  61. 37/38 refs: make lock genericMichael Haggerty, Jun 3, 2016
  62. Junio C HamanoJun 7, 2016
  63. Michael HaggertyJun 9, 2016
  64. Michael HaggertyJun 9, 2016
  65. 38/38 refs: implement iteration over only per-worktree refsMichael Haggerty, Jun 3, 2016
  66. Duy NguyenJul 10, 2016
  67. Michael HaggertyJul 13, 2016

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.