image-layout: fix inconsistent description about external blob store

[AkihiroSuda]

Following sentences were conflicting:

• A blob, referenced with digest <alg>:<hex> (per descriptor), MUST have its content stored in a file under blobs/<alg>/<hex>.
• The blobs directory MAY be missing referenced blobs, in which case the missing blobs SHOULD be fulfilled by an external blob store.

Signed-off-by: Akihiro Suda suda.akihiro@lab.ntt.co.jp

[wking]

I think that's weakening this too much, since it allows you to store <alg1>:<hex1> at blobs/<alg2>/<hex2> (or anywhere really). How about:

The content of blobs/<alg>/<hex> MUST match the digest <alg>:<hex>.

[AkihiroSuda]

done

[wking]

I'd rather not use <encoded> here, since we currently require hex there. If/when #654 lands, it can update the <hex> reference.

It would also be nice to use a valid descriptor like sha256:da39a3ee5e6b4b0d3255bfef95601890afd80709, although since the current master uses sha256:5b, that would be a somewhat orthogonal fixup (but it seems like a fairly straightforward fix, so I thought I'd mention it in case you want to pick it up here).

[AkihiroSuda]

#654 will land soon.
Even if #654 gets rejected, I don't think using <encoded> hurts here

[wking]

“<hex> and <encoded>” should be “<alg> and <hex>” (or “<alg> and <encoded>” in the post-#654 world).

[AkihiroSuda]

done

[AkihiroSuda]

rebased PTAL @stevvooe

[stevvooe]

LGTM

@AkihiroSuda Thank you, again!

[AkihiroSuda]

@vbatts PTAL?

[vbatts]

oh interesting. So this will have to match the character set being discussed for URI charset

[wking]

So this will have to match the character set being discussed for URI charset

I think you're talking about #671 for “URI charset”, but #671 is about org.opencontainers.image.ref.name while this PR is about digests. The digest charset is still fairly restricted.

[vbatts]

oh you're right. I'm confused.

…

On Tue, May 16, 2017 at 4:50 PM W. Trevor King ***@***.***> wrote: ***@***.**** commented on this pull request. ------------------------------ In image-layout.md <#656 (comment)> : > @@ -53,9 +53,8 @@ afff3924849e458c5ef237db5f89539274d5e609db5db935ed3959c90f1f 2d51 ./blobs/sha256/ ## Blobs * Object names in the `blobs` subdirectories are composed of a directory for each hash algorithm, the children of which will contain the actual content. -* A blob, referenced with digest `<alg>:<encoded>` (per [descriptor](descriptor.md#digests-and-verification)), MUST have its content stored in a file under `blobs/<alg>/<encoded>`. -* The character set of the entry name for `<encoded>` and `<alg>` MUST match the respective grammar elements described in [descriptor](descriptor.md#digests-and-verification). -* For example `sha256:5b` will map to the layout `blobs/sha256/5b`. +* The content of `blobs/<alg>/<encoded>` MUST match the digest `<alg>:<encoded>` (referenced per [descriptor](descriptor.md#digests-and-verification)). For example, the content of `blobs/sha256/da39a3ee5e6b4b0d3255bfef95601890afd80709` MUST match the digest `sha256:da39a3ee5e6b4b0d3255bfef95601890afd80709`. So this will have to match the character set being discussed for URI charset I think you're talking about #671 <#671> for “URI charset”, but #671 <#671> is about org.opencontainers.image.ref.name while this PR is about digests. The charset there is still fairly restricted <https://github.com/opencontainers/image-spec/blob/f03dbe35d449c54915d235f1a3cf8f585a24babe/descriptor.md#digests> . — You are receiving this because you were mentioned. Reply to this email directly, view it on GitHub <#656 (comment)>, or mute the thread <https://github.com/notifications/unsubscribe-auth/AAEF6U-6nP0GlqLilGDA72hf69JFKpCSks5r6gv2gaJpZM4NIRwk> .

[AkihiroSuda]

LGTY? 😄

[vbatts]

LGTM