Re: [PATCH v3 2/2] archive: remove USE_THE_REPOSITORY_VARIABLE from archival modules
- From
- Srujith Poojary <poojarysrujith@gmail.com>
- Date
- Nov 11, 2026, 08:34 UTC
- Message-ID
- <CABh_KWj31AHp9qvWxOEAN_T9xUkkP0b_P+0BxxU4ZNNeMOmNjQ@mail.gmail.com>
- In-Reply-To
- <xmqqse18pgin.fsf@gitster.g>
On Wed, Nov 11, 2026 at 2:10 AM Junio C Hamano <gitster@pobox.com> wrote:
Show 90 quoted lines
>
> "Srujith A Poojary via GitGitGadget" <gitgitgadget@gmail.com>
> writes:
>
> > From: Srujith A Poojary <poojarysrujith@gmail.com>
> >
> > Refactor archive-tar.c, archive-zip.c, archive.c, and supporting files
> > to clean up archiver initialization functions and header dependencies.
> >
> > Signed-off-by: Srujith A Poojary <poojarysrujith@gmail.com>
> > ---
> > archive-tar.c | 5 ++---
> > builtin/archive.c | 1 +
> > 2 files changed, 3 insertions(+), 3 deletions(-)
>
> Where is the change to archive-zip.c the proposed log message
> advertises?
>
> The proposed log message claims that it refactors and cleans up, but
> the patch looks like a totally incoherent and unexplained mess. All
> the places I note "What is this change about?" are unexplained
> changes whose value is dubious.
>
> > diff --git a/archive-tar.c b/archive-tar.c
> > index a1c66024d4..642a3ddfc3 100644
> > --- a/archive-tar.c
> > +++ b/archive-tar.c
> > @@ -3,7 +3,6 @@
> > */
> >
> > #define USE_THE_REPOSITORY_VARIABLE
> > -
> > #include "git-compat-util.h"
> > #include "config.h"
> > #include "gettext.h"
>
> What is this change about?
>
> > @@ -337,7 +336,7 @@ static void write_global_extended_header(struct archiver_args *args)
> > if (oid)
> > strbuf_append_ext_header(&ext_header, "comment",
> > oid_to_hex(oid),
> > - the_hash_algo->hexsz);
> > + args->repo->hash_algo->hexsz);
> > if (args->time > USTAR_MAX_MTIME) {
> > strbuf_append_ext_header_uint(&ext_header, "mtime",
> > args->time);
>
> Here, the args->repo which was relayed by archive.c:write_archive()
> is used. This function, when called by builtin/archive.c, passes
> the repo it obtained from git.c's dispatcher; when it is called by
> builtin/upload-archive.c, it uses the_repository (which is fine).
>
> This hunk looks like a good change.
>
> > @@ -529,13 +528,13 @@ static struct archiver tar_archiver = {
> > void init_tar_archiver(void)
> > {
> > int i;
> > + repo_config(the_repository, git_tar_config, NULL);
> > register_archiver(&tar_archiver);
> >
> > tar_filter_config("tar.tgz.command", internal_gzip_command, NULL);
> > tar_filter_config("tar.tgz.remote", "true", NULL);
> > tar_filter_config("tar.tar.gz.command", internal_gzip_command, NULL);
> > tar_filter_config("tar.tar.gz.remote", "true", NULL);
> > - repo_config(the_repository, git_tar_config, NULL);
> > for (i = 0; i < nr_tar_filters; i++) {
> > /* omit any filters that never had a command configured */
> > if (tar_filters[i]->filter_command)
>
> What is this change about? Earlier, we first declared the built-in
> default configurations and then read from the configuration files.
> Now you read the configuration files, and then overwrite the values
> the configuration files specified with our built-in default?
>
> > diff --git a/builtin/archive.c b/builtin/archive.c
> > index 3c1288a123..8b7d56f8cb 100644
> > --- a/builtin/archive.c
> > +++ b/builtin/archive.c
> > @@ -2,6 +2,7 @@
> > * Copyright (c) 2006 Franck Bui-Huu
> > * Copyright (c) 2006 Rene Scharfe
> > */
> > +#define USE_THE_REPOSITORY_VARIABLE
> > #include "builtin.h"
> > #include "archive.h"
> > #include "gettext.h"
>
> What is this change about?Hi Junio,
Thank you for the guidance on the configuration initialization order.
I have updated the patch series locally to address your feedback: 1. Restored the correct initialization order in init_tar_archiver() within archive-tar.c so built-in defaults are assigned first, followed by repo_config() to allow user/repository configurations to take precedence. 2. Cleaned up the branch history to drop an accidental submodule commit, keeping the PR isolated exclusively to the archive refactor. 3. Re-ran the test suite (t5000-tar-tree.sh and t5004-archive-corner-cases.sh), and all tests now pass successfully (90/90 on tar-tree).
Best regards, Srujith A Poojary