From: Michael Forney <mforney@mforney.org>
To: 9front@9front.org
Subject: Re: [9front] git/pull: fetch all branches (please test)
Date: Tue, 19 Apr 2022 12:31:31 -0700 [thread overview]
Message-ID: <CAGw6cBsezD_G=xbvWBWGGUn5+Z_ZHbBf=xzPyN=ALoYBuqAKOQ@mail.gmail.com> (raw)
In-Reply-To: <404C9B5CFE08FE9146154E6D8964C191@eigenstate.org>
On 2022-04-11, ori@eigenstate.org <ori@eigenstate.org> wrote:
> diff 1d9d4ffef89e883949261ec9c56c57e0344726d7 uncommitted
> --- a/sys/src/cmd/git/fetch.c
> +++ b/sys/src/cmd/git/fetch.c
> @@ -180,12 +180,13 @@
> int
> fetchpack(Conn *c)
> {
> - char buf[Pktmax], *sp[3];
> + char buf[Pktmax], *sp[3], *ep;
> char *packtmp, *idxtmp, **ref;
> Hash h, *have, *want;
> int nref, refsz, first;
> - int i, n, req, pfd;
> + int i, n, l, req, pfd;
> vlong packsz;
> + Objset hadobj;
> Object *o;
>
> nref = 0;
> @@ -246,13 +247,19 @@
> req = 1;
> }
> flushpkt(c);
> + osinit(&hadobj);
> for(i = 0; i < nref; i++){
> - if(hasheq(&have[i], &Zhash))
> + if(hasheq(&have[i], &Zhash) || oshas(&hadobj, have[i]))
> continue;
> + if((o = readobject(have[i])) == nil)
> + sysfatal("missing object we should have: %H", have[i]);
> + osadd(&hadobj, o);
> + unref(o);
> n = snprint(buf, sizeof(buf), "have %H\n", have[i]);
> if(writepkt(c, buf, n + 1) == -1)
> sysfatal("could not send have for %H", have[i]);
> }
> + osclear(&hadobj);
I've been looking over the http-protocol docs, and it describes an
algorithm involving a priority queue and commit graph traversal that
we don't seem to do.
How does this work when our refs have several local commits that the
server doesn't have?
> if(!req)
> flushpkt(c);
>
> @@ -260,7 +267,7 @@
> if(writepkt(c, buf, n) == -1)
> sysfatal("write: %r");
> if(!req)
> - return 0;
> + goto showrefs;
> if(readphase(c) == -1)
> sysfatal("read: %r");
> if((n = readpkt(c, buf, sizeof(buf))) == -1)
> @@ -277,7 +284,28 @@
> sysfatal("could not create %s: %r", packtmp);
>
> fprint(2, "fetching...\n");
> - packsz = 0;
> + /*
> + * Work around torvalds git bug: we get duplicate have lines
Do you mean duplicate ACK lines? From my reading of the protocol docs,
"have" lines are only sent by us (the client).
In fact, I don't really see much about ACK lines at all. Do you know
if this is described anywhere in the git docs?
> + * somtimes, even though the protocol is supposed to start the
Typo here.
> + * pack file immediately.
> + *
> + * Skip ahead until we read 'PACK' off the wire
> + */
> + while(1){
> + if(readn(c->rfd, buf, 4) != 4)
> + sysfatal("fetch packfile: short read");
> + buf[4] = 0;
> + if(strncmp(buf, "PACK", 4) == 0)
> + break;
> + l = strtol(buf, &ep, 16);
> + if(l == 0 || ep != buf + 4)
> + sysfatal("fetch packfile: junk pktline");
> + if(readn(c->rfd, buf, l) != l)
> + sysfatal("fetch packfile: short read");
> + }
> + if(write(pfd, "PACK", 4) != 4)
> + sysfatal("write pack header: %r");
> + packsz = 4;
> while(1){
> n = read(c->rfd, buf, sizeof buf);
> if(n == 0)
> --- a/sys/src/cmd/git/pull
> +++ b/sys/src/cmd/git/pull
> @@ -7,13 +7,10 @@
> upstream=$2
> url=$3
> dir=$4
> - bflag=()
> dflag=()
> - if(! ~ $#branch 0)
> - bflag=(-b $branch)
> if(! ~ $#debug 0)
> dflag='-d'
> - {git/fetch $dflag $bflag -u $upstream $url >[2=3] || die $status} | awk '
> + {git/fetch $dflag -u $upstream $url >[2=3] || die $status} | awk '
> /^remote/{
> if($2=="HEAD")
> next
Now that $branch is unused, I think you can remove the corresponding
positional parameter in the update function, as well as the git/pull
-a and -b options.
next prev parent reply other threads:[~2022-04-19 19:33 UTC|newest]
Thread overview: 5+ messages / expand[flat|nested] mbox.gz Atom feed top
2022-04-11 3:19 ori
2022-04-11 21:32 ` ori
2022-04-12 2:55 ` ori
2022-04-19 19:31 ` Michael Forney [this message]
2022-04-20 3:42 ` ori
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
* Reply using the --to, --cc, and --in-reply-to
switches of git-send-email(1):
git send-email \
--in-reply-to='CAGw6cBsezD_G=xbvWBWGGUn5+Z_ZHbBf=xzPyN=ALoYBuqAKOQ@mail.gmail.com' \
--to=mforney@mforney.org \
--cc=9front@9front.org \
/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.
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 NNTP newsgroup(s).