From: Jeff Hostetler <git@jeffhostetler.com>
To: Jonathan Tan <jonathantanmy@google.com>
Cc: git@vger.kernel.org, peartben@gmail.com,
Christian Couder <christian.couder@gmail.com>
Subject: Re: RFC: Design and code of partial clones (now, missing commits and trees OK) (part 2/3)
Date: Fri, 22 Sep 2017 17:19:50 -0400 [thread overview]
Message-ID: <5b251b02-fe00-3d99-8ae2-d6df0193ebe0@jeffhostetler.com> (raw)
In-Reply-To: <20170921155150.1d57d89e@twelve2.svl.corp.google.com>
On 9/21/2017 6:51 PM, Jonathan Tan wrote:
> On Thu, 21 Sep 2017 13:59:43 -0400
> Jeff Hostetler <git@jeffhostetler.com> wrote:
>
>> (part 2)
>>
>> Additional overall comments on:
>> https://github.com/jonathantanmy/git/commits/partialclone2
>>
>> {} I think it would help to split the blob-max-bytes filtering and the
>> promisor/promised concepts and discuss them independently.
>>
>> {} Then we can talk about about the promisor/promised
>> functionality independent of any kind of filter. The net-net is that
>> the client has missing objects and it doesn't matter what filter
>> criteria or mechanism caused that to happened.
>>
>> {} blob-max-bytes is but one such filter we should have. This
>> might be sufficient if the goal is replace LFS (where you rarely ever
>> need any given very very large object) and dynamically loading
>> them as needed is sufficient and the network round-trip isn't
>> too much of a perf penalty.
>>
>> {} But if we want to do things like a "sparse-enlistments" where
>> the client only needs a small part of the tree using sparse-checkout.
>> For example, only populating 50,000 files from a tree of 3.5M
>> files at HEAD, then we need a more general filtering.
>>
>> {} And as I said above, how we chose to filter should be
>> independent of how the client handles promisor/promised objects.
>
> I agree that they are independent. (I put everything together so that
> people could see how they work together, but they can be changed
> independently of each other.)
>
>> {} Also, if we rely strictly on dynamic object fetching to fetch
>> missing objects, we are effectively tethered to the server during
>> operations (such as checkout) that the user might not think about as
>> requiring a network connection. And we are forced to keep the same
>> limitations of LFS in that you can't prefetch and go offline (without
>> actually checking out to your worktree first). And we can't bulk or
>> parallel fetch objects.
>
> I don't think dynamic object fetching precludes any other more optimized
> way of fetching or prefetching - I implemented dynamic object fetching
> first so that we would have a fallback, but the others definitely can be
> implemented too.
yes, we need that as a fallback/default for the odd cases where we
can't predict perfectly. Like during a blame or history or a merge.
I didn't mean to say we didn't need it, but rather that we should
try to minimize it.
>
>> {} I think it would also help to move the blob-max-bytes calculation
>> out of pack-objects.c : add_object_entry() [1]. The current code
>> isolates the computation there so that only pack-objects can do the
>> filtering.
>>
>> Instead, put it in list-objects.c and traverse_commit_list() so
>> that pack-objects and rev-list can share it (as Peff suggested [2] in
>> response to my first patch series in March).
>>
>> For example, this would let the client have a pre-checkout hook,
>> use rev-list to compute the set of missing objects needed for that
>> commit, and pipe that to a command to BULK fetch them from the server
>> BEFORE starting the actual checkout. This would allow the savy user
>> to manually run a prefetch before going offline.
>>
>> [1]
>> https://github.com/jonathantanmy/git/commit/68e529484169f4800115c5a32e0904c25ad14bd8#diff-a8d2c9cf879e775d748056cfed48440cR1110
>>
>> [2]
>> https://public-inbox.org/git/20170309073117.g3br5btsfwntcdpe@sigill.intra.peff.net/
>
> In your specific example, how would rev-list know, on the client, to
> include (or exclude) a large blob in its output if it does not have it,
> and thus does not know its size?
>
The client doesn't have the size. It just knows it is missing and it
needs it. It doesn't matter why it is missing. (But I guess the client
could assume it is because it is large.)
So rev-list on the client could filter the objects it has by size.
I added that to rev-list primarily to demonstrate and debug the
filtering concept (it's easier than playing with packfiles). But
it can be used to drive client-side queries and bulk requests.
> My reason for including it in pack-objects.c is because I only needed it
> there and it is much simpler, but I agree that if it can be used
> elsewhere, we can put it in a more general place.
>
>> {} This also locks us into size-only filtering and makes it more
>> difficult to add other filters. In that the add_object_entry()
>> code gets called on an object after the traversal has decided
>> what to do with it. It would be difficult to add tree-trimming
>> at this level, for example.
>
> That is true.
>
>> {} An early draft of this type of filtering is here [3]. I hope to
>> push up a revised draft of this shortly.
>>
>> [3]
>> https://public-inbox.org/git/20170713173459.3559-1-git@jeffhostetler.com/
>
> OK - I'll take a look when that is done (I think I commented on an
> earlier version on that).
>
FYI I just posted my RFC this afternoon.
https://public-inbox.org/git/20170922204211.GA24036@google.com/T/
Thanks
Jeff
next prev parent reply other threads:[~2017-09-22 21:19 UTC|newest]
Thread overview: 21+ messages / expand[flat|nested] mbox.gz Atom feed top
2017-09-15 20:43 RFC: Design and code of partial clones (now, missing commits and trees OK) Jonathan Tan
2017-09-19 5:51 ` Junio C Hamano
2017-09-21 17:57 ` Jeff Hostetler
2017-09-21 22:42 ` Jonathan Tan
2017-09-22 21:02 ` Jeff Hostetler
2017-09-22 22:49 ` Jonathan Tan
2017-09-26 15:26 ` Michael Haggerty
2017-09-29 20:21 ` Jonathan Tan
2017-09-21 17:59 ` RFC: Design and code of partial clones (now, missing commits and trees OK) (part 2/3) Jeff Hostetler
2017-09-21 22:51 ` Jonathan Tan
2017-09-22 21:19 ` Jeff Hostetler [this message]
2017-09-22 22:52 ` Jonathan Tan
2017-09-26 14:03 ` Jeff Hostetler
2017-09-21 18:00 ` RFC: Design and code of partial clones (now, missing commits and trees OK) (part 3) Jeff Hostetler
2017-09-21 23:04 ` Jonathan Tan
2017-09-22 21:32 ` Jeff Hostetler
2017-09-22 22:58 ` Jonathan Tan
2017-09-26 14:25 ` Jeff Hostetler
2017-09-26 17:32 ` Jonathan Tan
2017-09-29 0:53 ` RFC: Design and code of partial clones (now, missing commits and trees OK) Jonathan Tan
2017-09-29 2:03 ` Junio C Hamano
Reply instructions:
You may reply publicly to this message via plain-text email
using any one of the following methods:
* Save the following mbox file, import it into your mail client,
and reply-to-all from there: mbox
Avoid top-posting and favor interleaved quoting:
https://en.wikipedia.org/wiki/Posting_style#Interleaved_style
List information: http://vger.kernel.org/majordomo-info.html
* Reply using the --to, --cc, and --in-reply-to
switches of git-send-email(1):
git send-email \
--in-reply-to=5b251b02-fe00-3d99-8ae2-d6df0193ebe0@jeffhostetler.com \
--to=git@jeffhostetler.com \
--cc=christian.couder@gmail.com \
--cc=git@vger.kernel.org \
--cc=jonathantanmy@google.com \
--cc=peartben@gmail.com \
/path/to/YOUR_REPLY
https://kernel.org/pub/software/scm/git/docs/git-send-email.html
* If your mail client supports setting the In-Reply-To header
via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line
before the message body.
Code repositories for project(s) associated with this public inbox
https://80x24.org/mirrors/git.git
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for read-only IMAP folder(s) and NNTP newsgroup(s).