Re: [PATCH][GSoC] submodule: introduce add-clone helper for submodule add

[Atharva Raykar <raykar.ath@gmail.com>]

On 02-Jun-2021, at 03:40, Christian Couder <christian.couder@gmail.com> wrote:
> 
> On Fri, May 28, 2021 at 10:13 AM Atharva Raykar <raykar.ath@gmail.com> wrote:
>> 
>> Convert the the shell code that performs the cloning of the repository that is
> 
> s/the the/the/
> 
>> to be added, and checks out to the appropriate branch.
> 
> Something a bit more explicit might make things easier to understand.
> For example:
> 
> "Let's add a new "add-clone" subcommand to `git submodule--helper`
> with the goal of converting part of the shell code in git-submodule.sh
> related to `git submodule add` into C code. This new subcommand clones
> the repository that is to be added, and checks out to the appropriate
> branch."
> 
> Then a simpler title could be:
> 
> "submodule--helper: introduce add-clone subcommand"

Great suggestions. I'll update my commit message.

>> This is meant to be a faithful conversion that leaves the behaviour of
>> 'submodule add' unchanged. The only minor change is that if a submodule name has
>> been supplied with a name that clashes with a local submodule, the message shown
>> to the user ("A git directory for 'foo' is found locally...") is prepended with
>> "error" for clarity.
> 
> Good.
> 
>> Signed-off-by: Atharva Raykar <raykar.ath@gmail.com>
>> Mentored-by: Christian Couder <christian.couder@gmail.com>
>> Mentored-by: Shourya Shukla <shouryashukla.oo@gmail.com>
>> Based-on-patch-by: Shourya Shukla <shouryashukla.oo@gmail.com>
>> Based-on-patch-by: Prathamesh Chavan <pc44800@gmail.com>
>> ---
>> 
>> This is part of a series of changes that will result in all of 'submodule add'
>> being converted to C, which is a more familiar language for Git developers, and
>> paves the way to improve performance and portability.
>> 
>> I have made this patch based on Shourya's patch[1]. I have decided to send the
>> changes in smaller, more reviewable parts. The add-clone subcommand of
>> submodule--helper is an intermediate change, while I work on translating all of
>> the code. So in the next few patches, this helper subcommand is likely to be
>> removed as its functionality would be invoked from the C code itself.
> 
> It might be a good idea to let us know how many such new subcommands
> you'd like to introduce before removing them.

I'll add that in my description of v2.

> Anyway I think it's a good idea to send changes in smaller, more
> easily reviewable parts. Hopefully this way more work will end up
> being merged.
> 
>> [...]
>> +static void show_fetch_remotes(FILE *output, const char *sm_name, const char *git_dir_path)
>> +{
>> +       struct child_process cp_remote = CHILD_PROCESS_INIT;
>> +       struct strbuf sb_remote_out = STRBUF_INIT;
>> +
>> +       cp_remote.git_cmd = 1;
>> +       strvec_pushf(&cp_remote.env_array,
>> +                    "GIT_DIR=%s", git_dir_path);
>> +       strvec_push(&cp_remote.env_array, "GIT_WORK_TREE=.");
>> +       strvec_pushl(&cp_remote.args, "remote", "-v", NULL);
>> +       if (!capture_command(&cp_remote, &sb_remote_out, 0)) {
>> +               char *next_line, *name, *url, *tail;
> 
> Maybe name, url and tail could be declared in the while loop below
> where they are used.

Will do. Just to better understand your intent, is the reason to
do this to make the declarations closer to usage, for the sake of
better readability?

I've not yet fully developed a taste for good C style, so I wanted
to ask, which one looks better to you in these?

/* Sample 1 */
while (begin != end && (line = get_next_line(begin, end))) {
	char *name, *url, *tail;
	name = parse_token(&begin, next_line);
	url = parse_token(&begin, next_line);
	tail = parse_token(&begin, next_line);
	...
}

/* Sample 2 */
while (begin != end && (line = get_next_line(begin, end))) {
	char *name = parse_token(&begin, next_line);
	char *url = parse_token(&begin, next_line);
	char *tail = parse_token(&begin, next_line);
	...
}

>> +               char *begin = sb_remote_out.buf;
>> +               char *end = sb_remote_out.buf + sb_remote_out.len;
>> +               while (begin != end &&
>> +                      (next_line = get_next_line(begin, end))) {
> 
> It would be nice if the above 2 lines could be reduced into just one
> line. Maybe renaming "next_line" to just "line" could help with that.

Noted.

>> +                       name = parse_token(&begin, next_line);
>> +                       url = parse_token(&begin, next_line);
>> +                       tail = parse_token(&begin, next_line);
>> +                       if (!memcmp(tail, "(fetch)", 7))
>> +                               fprintf(output, "  %s\t%s\n", name, url);
>> +                       free(url);
>> +                       free(name);
>> +                       free(tail);
>> +               }
>> +       }
>> +
>> +       strbuf_release(&sb_remote_out);
>> +}
>> +
>> +static int add_submodule(const struct add_data *info)
>> +{
>> +       char *submod_gitdir_path;
>> +       /* perhaps the path already exists and is already a git repo, else clone it */
>> +       if (is_directory(info->sm_path)) {
>> +               printf("sm_path=%s\n", info->sm_path);
> 
> Is this a leftover debug statement?

Nope, at least not _my_ leftover debug statement.

I saw it in git-submodule.sh here, so I preserved it:

-	# perhaps the path exists and is already a git repo, else clone it
-	if test -e "$sm_path"
-	...

Personally, I found that comment quite useful when I was trying to
understand the shell version, because at a glance I immediately
knew what the intention of the big block of code was, that followed
the comment.

Perhaps it could be broken into many functions to make it more
readable without needing a comment, but that is outside the scope
of this particular patch, which is aiming for a faithful conversion.

>> +               submod_gitdir_path = xstrfmt("%s/.git", info->sm_path);
>> +               if (is_directory(submod_gitdir_path) || file_exists(submod_gitdir_path))
>> +                       printf(_("Adding existing path at '%s' to index\n"),
>> +                              info->sm_path);
>> +               else
>> +                       die(_("'%s' already exists and is not a valid git repo"),
>> +                           info->sm_path);
>> +               free(submod_gitdir_path);
>> +       } else {
>> +               struct strvec clone_args = STRVEC_INIT;
>> +               struct child_process cp = CHILD_PROCESS_INIT;
>> +               submod_gitdir_path = xstrfmt(".git/modules/%s", info->sm_name);
>> +
>> +               if (is_directory(submod_gitdir_path)) {
>> +                       if (!info->force) {
>> +                               error(_("A git directory for '%s' is found "
>> +                                       "locally with remote(s):"), info->sm_name);
>> +                               show_fetch_remotes(stderr, info->sm_name,
>> +                                                  submod_gitdir_path);
>> +                               fprintf(stderr,
>> +                                       _("If you want to reuse this local git "
>> +                                         "directory instead of cloning again from\n"
>> +                                         "  %s\n"
>> +                                         "use the '--force' option. If the local git "
>> +                                         "directory is not the correct repo\n"
>> +                                         "or if you are unsure what this means, choose "
>> +                                         "another name with the '--name' option.\n"),
>> +                                       info->realrepo);
>> +                               free(submod_gitdir_path);
>> +                               return 1;
>> +                       } else {
>> +                               printf(_("Reactivating local git directory for "
>> +                                        "submodule '%s'\n"), info->sm_name);
>> +                       }
>> +               }
>> +               free(submod_gitdir_path);
>> +
>> +               strvec_push(&clone_args, "clone");
>> +
>> +               if (info->quiet)
>> +                       strvec_push(&clone_args, "--quiet");
>> +
>> +               if (info->progress)
>> +                       strvec_push(&clone_args, "--progress");
>> +
>> +               if (info->prefix)
>> +                       strvec_pushl(&clone_args, "--prefix", info->prefix, NULL);
>> +
>> +               strvec_pushl(&clone_args, "--path", info->sm_path, "--name",
>> +                            info->sm_name, "--url", info->realrepo, NULL);
> 
> Maybe this unconditional strvec_pushl(...) could be squashed into the
> strvec_push(&clone_args, "clone") above.

Got it.

>> +               if (info->reference_path)
>> +                       strvec_pushl(&clone_args, "--reference",
>> +                                    info->reference_path, NULL);
>> +
>> +               if (info->dissociate)
>> +                       strvec_push(&clone_args, "--dissociate");
>> +
> 
> Blank lines since the above strvec_push(&clone_args, "clone") could
> perhaps be removed.

Will do.

>> [...]
>> +       const char *const usage[] = {
>> +               N_("git submodule--helper clone [--prefix=<path>] [--quiet] [--force] "
> 
> s/clone/add-clone/
> 
>> +                  "[--reference <repository>] [--depth <depth>] [-b|--branch <branch>]"
>> +                  "--url <url> --path <path> --name <name>"),
> 
> The --progress and --dissociate options seem to be missing.

Thanks, will fix.

>> +               NULL
>> +       };