threads / patch / 66566

patchmerge-blobs: stop using the_repository

Subject: [Outreachy PATCH] merge-blobs: stop using the_repository

## tl;dr

6 messages between Oct 17, 2026 and Nov 7, 2026. Diffs are folded; open one to read it.

replies: 5people: 3as markdown or json

Okhuomon Ajayi· Oct 17, 2026, 10:52 UTC · lore

merge_blobs() already has the repository through istate. Pass it to fill_mmfile_blob() and use it when reading objects.

Remove the USE_THE_REPOSITORY_VARIABLE definition and include the headers needed for the structure definitions.

Signed-off-by: Okhuomon Ajayi <okhuomonajayi54@gmail.com>
---
 merge-blobs.c | 17 +++++++++--------
 1 file changed, 9 insertions(+), 8 deletions(-)
Show changes to merge-blobs.c +9 −8
diff --git a/merge-blobs.c b/merge-blobs.c
index 49dbec9529..1b0621c814 100644
--- a/merge-blobs.c
+++ b/merge-blobs.c
@@ -1,18 +1,19 @@
-#define USE_THE_REPOSITORY_VARIABLE
-
 #include "git-compat-util.h"
 #include "merge-ll.h"
 #include "blob.h"
 #include "merge-blobs.h"
 #include "odb.h"
+#include "read-cache-ll.h"
+#include "repository.h"
 
-static int fill_mmfile_blob(mmfile_t *f, struct blob *obj)
+static int fill_mmfile_blob(struct repository *r, mmfile_t *f,
+			    struct blob *obj)
 {
 	void *buf;
 	size_t size;
 	enum object_type type;
 
-	buf = odb_read_object(the_repository->objects, &obj->object.oid,
+	buf = odb_read_object(r->objects, &obj->object.oid,
 			      &type, &size);
 	if (!buf)
 		return -1;
@@ -79,17 +80,17 @@ void *merge_blobs(struct index_state *istate, const char *path,
 			return NULL;
 		if (!our)
 			our = their;
-		return odb_read_object(the_repository->objects, &our->object.oid,
+		return odb_read_object(istate->repo->objects, &our->object.oid,
 				       &type, size);
 	}
 
-	if (fill_mmfile_blob(&f1, our) < 0)
+	if (fill_mmfile_blob(istate->repo, &f1, our) < 0)
 		goto out_no_mmfile;
-	if (fill_mmfile_blob(&f2, their) < 0)
+	if (fill_mmfile_blob(istate->repo, &f2, their) < 0)
 		goto out_free_f1;
 
 	if (base) {
-		if (fill_mmfile_blob(&common, base) < 0)
+		if (fill_mmfile_blob(istate->repo, &common, base) < 0)
 			goto out_free_f2_f1;
 	} else {
 		common.ptr = xstrdup("");
-- 
2.54.0
Usman Akinyemi· Oct 20, 2026, 16:53 UTC · re: Okhuomon Ajayi · lore

Re: [Outreachy PATCH] merge-blobs: stop using the_repository

Show 9 quoted lines
>
> merge_blobs() already has the repository through istate.
> Pass it to fill_mmfile_blob() and use it when reading objects.
>
> Remove the USE_THE_REPOSITORY_VARIABLE definition and
> include the headers needed for the structure definitions.
>
> Signed-off-by: Okhuomon Ajayi <okhuomonajayi54@gmail.com>
> ---
Good.
Show 21 quoted lines
>  merge-blobs.c | 17 +++++++++--------
>  1 file changed, 9 insertions(+), 8 deletions(-)
>
> diff --git a/merge-blobs.c b/merge-blobs.c
> index 49dbec9529..1b0621c814 100644
> --- a/merge-blobs.c
> +++ b/merge-blobs.c
> @@ -1,18 +1,19 @@
> -#define USE_THE_REPOSITORY_VARIABLE
> -
>  #include "git-compat-util.h"
>  #include "merge-ll.h"
>  #include "blob.h"
>  #include "merge-blobs.h"
>  #include "odb.h"
> +#include "read-cache-ll.h"
> +#include "repository.h"
>
> -static int fill_mmfile_blob(mmfile_t *f, struct blob *obj)
> +static int fill_mmfile_blob(struct repository *r, mmfile_t *f,
> +                           struct blob *obj)
Is there a reason why the repo variable is named "r" instead of repo ?

"repo" seems more "descriptive" than "r". if there is no reason, let's change it to "repo" from "r"

Show 8 quoted lines
>  {
>         void *buf;
>         size_t size;
>         enum object_type type;
>
> -       buf = odb_read_object(the_repository->objects, &obj->object.oid,
> +       buf = odb_read_object(r->objects, &obj->object.oid,
>                               &type, &size);
this r->objects will become repo->objects
Show 25 quoted lines
>         if (!buf)
>                 return -1;
> @@ -79,17 +80,17 @@ void *merge_blobs(struct index_state *istate, const char *path,
>                         return NULL;
>                 if (!our)
>                         our = their;
> -               return odb_read_object(the_repository->objects, &our->object.oid,
> +               return odb_read_object(istate->repo->objects, &our->object.oid,
>                                        &type, size);
>         }
>
> -       if (fill_mmfile_blob(&f1, our) < 0)
> +       if (fill_mmfile_blob(istate->repo, &f1, our) < 0)
>                 goto out_no_mmfile;
> -       if (fill_mmfile_blob(&f2, their) < 0)
> +       if (fill_mmfile_blob(istate->repo, &f2, their) < 0)
>                 goto out_free_f1;
>
>         if (base) {
> -               if (fill_mmfile_blob(&common, base) < 0)
> +               if (fill_mmfile_blob(istate->repo, &common, base) < 0)
>                         goto out_free_f2_f1;
>         } else {
>                 common.ptr = xstrdup("");
> --
apart from the nit about r vs repo this looks good.
> 2.54.0
>
Thanks.
Okhuomon Ajayi· Oct 20, 2026, 21:41 UTC · re: Usman Akinyemi · lore

Re: [Outreachy PATCH] merge-blobs: stop using the_repository

On Tue, Oct 20, 2026 at 5:53 PM Usman Akinyemi <usmanakinyemi202@gmail.com> wrote:

> Is there a reason why the repo variable is named "r" instead of repo ?
>
> "repo" seems more "descriptive" than "r". if there is no reason, let's change
> it to "repo" from "r"

Thanks for the review. There is no particular reason. I studied some files and saw it written as "r", but I will change it to "repo" in v2 so it is more descriptive, as you said. Okhuomon

Junio C Hamano· Oct 20, 2026, 21:54 UTC · re: Usman Akinyemi · lore

Re: [Outreachy PATCH] merge-blobs: stop using the_repository

Usman Akinyemi <usmanakinyemi202@gmail.com> writes:
Show 7 quoted lines
>> +static int fill_mmfile_blob(struct repository *r, mmfile_t *f,
>> +                           struct blob *obj)
>
> Is there a reason why the repo variable is named "r" instead of repo ?
>
> "repo" seems more "descriptive" than "r". if there is no reason, let's change
> it to "repo" from "r"
Is that really needed?

I thought that we already use short-and-sweet "r" elsewhere, and it is expected we will have many more functions that take "struct repository *" as the first parameter as the libification stuff goes further, so if we consistently say "r", that would be just familiar and descriptive as "repository" to our developers, without risking to make readers wonder what "repo" stands for, things other than "repository", like "reposses" ;-).

Usman Akinyemi· Oct 20, 2026, 22:16 UTC · re: Junio C Hamano · lore

Re: [Outreachy PATCH] merge-blobs: stop using the_repository

Show 10 quoted lines
>
> Is that really needed?
>
> I thought that we already use short-and-sweet "r" elsewhere, and it
> is expected we will have many more functions that take "struct
> repository *" as the first parameter as the libification stuff goes
> further, so if we consistently say "r", that would be just familiar
> and descriptive as "repository" to our developers, without risking
> to make readers wonder what "repo" stands for, things other than
> "repository", like "reposses" ;-).

We also have "repo" used in some places, we might want to pick one, but I do not think it is a big issue.

Since, there instances of "r" Okhuomon does not have to send a new patch for changing the variable name then.

Thanks.
Junio C Hamano· Nov 7, 2026, 02:30 UTC · re: Okhuomon Ajayi · lore

Re: [Outreachy PATCH] merge-blobs: stop using the_repository

Okhuomon Ajayi <okhuomonajayi54@gmail.com> writes:
Show 10 quoted lines
> merge_blobs() already has the repository through istate.
> Pass it to fill_mmfile_blob() and use it when reading objects.
>
> Remove the USE_THE_REPOSITORY_VARIABLE definition and
> include the headers needed for the structure definitions.
>
> Signed-off-by: Okhuomon Ajayi <okhuomonajayi54@gmail.com>
> ---
>  merge-blobs.c | 17 +++++++++--------
>  1 file changed, 9 insertions(+), 8 deletions(-)

Even though I would have introduced a single repo variable instead of dereferencing istate->repo number of times inside merge_blobs() function, that probably falls into personal preference category, so insignificant a difference that once code is written in one way, it is not worth going in and rewrite into the other way.

Let's mark the topic as ready for 'next'.  Thanks.
Show 9 quoted lines
> -		return odb_read_object(the_repository->objects, &our->object.oid,
> +		return odb_read_object(istate->repo->objects, &our->object.oid,
> -	if (fill_mmfile_blob(&f1, our) < 0)
> +	if (fill_mmfile_blob(istate->repo, &f1, our) < 0)
>  		goto out_no_mmfile;
> -	if (fill_mmfile_blob(&f2, their) < 0)
> +	if (fill_mmfile_blob(istate->repo, &f2, their) < 0)
> -		if (fill_mmfile_blob(&common, base) < 0)
> +		if (fill_mmfile_blob(istate->repo, &common, base) < 0)

← back to recent threads