# [PATCH] git-clone: fetch possibly detached HEAD over dumb http 9 messages from 2007-06-28 to 2007-07-01. Participants: Sven Verdoolaege, Junio C Hamano, Alex Riesen. Thread: https://gitlist.dev/t/8757 ## Sven Verdoolaege, 2007-06-28 10:52 Subject: [PATCH] git-clone: fetch possibly detached HEAD over dumb http Message-ID: <20070628105208.GA11105@liacs.nl> URL: https://gitlist.dev/e/20070628105208.GA11105%40liacs.nl ``` git-clone supports cloning from a repo with detached HEAD, but if this HEAD is not behind any branch tip then it would not have been fetched over dumb http, resulting in a fatal: Not a valid object name HEAD Since 928c210a, this would also happen on a http repo with a HEAD that is a symbolic link where someone has forgotton to run update-server-info. Signed-off-by: Sven Verdoolaege --- git-clone.sh | 3 ++- 1 files changed, 2 insertions(+), 1 deletions(-) diff --git a/git-clone.sh b/git-clone.sh index bd44ce1..cdbbc20 100755 --- a/git-clone.sh +++ b/git-clone.sh @@ -70,7 +70,8 @@ Perhaps git-update-server-info needs to be run there?" git-http-fetch $v -a -w "$tname" "$sha1" "$1" || exit 1 done <"$clone_tmp/refs" rm -fr "$clone_tmp" - http_fetch "$1/HEAD" "$GIT_DIR/REMOTE_HEAD" || + http_fetch "$1/HEAD" "$GIT_DIR/REMOTE_HEAD" && + git-http-fetch $v -a $(cat "$GIT_DIR/REMOTE_HEAD") "$1" || rm -f "$GIT_DIR/REMOTE_HEAD" } -- 1.5.2.2.585.g9cc0-dirty ``` ## Junio C Hamano, 2007-06-29 00:02 Subject: Re: [PATCH] git-clone: fetch possibly detached HEAD over dumb http Message-ID: <7vsl8bmxv9.fsf@assigned-by-dhcp.pobox.com> URL: https://gitlist.dev/e/7vsl8bmxv9.fsf%40assigned-by-dhcp.pobox.com In-Reply-To: <20070628105208.GA11105@liacs.nl> ``` Sven Verdoolaege writes: > git-clone supports cloning from a repo with detached HEAD, > but if this HEAD is not behind any branch tip then it > would not have been fetched over dumb http, resulting in a > > fatal: Not a valid object name HEAD > > Since 928c210a, this would also happen on a http repo > with a HEAD that is a symbolic link where someone has > forgotton to run update-server-info. > > Signed-off-by: Sven Verdoolaege Ok. But I think the change regresses when the remote side is actually on a particular branch, and is using symref to represent $GIT_DIR/HEAD. > git-clone.sh | 3 ++- > 1 files changed, 2 insertions(+), 1 deletions(-) > > diff --git a/git-clone.sh b/git-clone.sh > index bd44ce1..cdbbc20 100755 > --- a/git-clone.sh > +++ b/git-clone.sh > @@ -70,7 +70,8 @@ Perhaps git-update-server-info needs to be run there?" > git-http-fetch $v -a -w "$tname" "$sha1" "$1" || exit 1 > done <"$clone_tmp/refs" > rm -fr "$clone_tmp" > - http_fetch "$1/HEAD" "$GIT_DIR/REMOTE_HEAD" || > + http_fetch "$1/HEAD" "$GIT_DIR/REMOTE_HEAD" && > + git-http-fetch $v -a $(cat "$GIT_DIR/REMOTE_HEAD") "$1" || > rm -f "$GIT_DIR/REMOTE_HEAD" > } At this point, "$GIT_DIR/REMOTE_HEAD" is a copy of HEAD obtained from the remote site via curl. It can contain: (1) raw SHA-1 of the tip commit, if the HEAD is detached, or the repository uses a symlink to represent HEAD, or (2) "ref: refs/heads/$currentbranch". You would want to do this extra fetch only in case (1). I think the additional fetch would fail in case (2), and result in removal of $GIT_DIR/REMOTE_HEAD. Hmm? ``` ## Sven Verdoolaege, 2007-06-29 08:11 Subject: Re: [PATCH] git-clone: fetch possibly detached HEAD over dumb http Message-ID: <20070629081100.GA7969MdfPADPa@greensroom.kotnet.org> URL: https://gitlist.dev/e/20070629081100.GA7969MdfPADPa%40greensroom.kotnet.org In-Reply-To: <7vsl8bmxv9.fsf@assigned-by-dhcp.pobox.com> ``` On Thu, Jun 28, 2007 at 05:02:18PM -0700, Junio C Hamano wrote: > You would want to do this extra fetch only in case (1). > I think the additional fetch would fail in case (2), and result > in removal of $GIT_DIR/REMOTE_HEAD. You're right. It looks like I only tested it on symbolic link HEADs. Sorry about that. Will send a corrected patch later. skimo ``` ## Sven Verdoolaege, 2007-06-29 08:31 Subject: [PATCH] git-clone: fetch possibly detached HEAD over dumb http Message-ID: <20070629083108.GA14747@liacs.nl> URL: https://gitlist.dev/e/20070629083108.GA14747%40liacs.nl In-Reply-To: <7vsl8bmxv9.fsf@assigned-by-dhcp.pobox.com> ``` git-clone supports cloning from a repo with detached HEAD, but if this HEAD is not behind any branch tip then it would not have been fetched over dumb http, resulting in a fatal: Not a valid object name HEAD Since 928c210a, this would also happen on a http repo with a HEAD that is a symbolic link where someone has forgotton to run update-server-info. Signed-off-by: Sven Verdoolaege --- On Thu, Jun 28, 2007 at 05:02:18PM -0700, Junio C Hamano wrote: > Ok. But I think the change regresses when the remote side is > actually on a particular branch, and is using symref to > represent $GIT_DIR/HEAD. Updated patch tested on both symbolic links and symrefs. skimo git-clone.sh | 11 +++++++++++ 1 files changed, 11 insertions(+), 0 deletions(-) diff --git a/git-clone.sh b/git-clone.sh index bd44ce1..4cbf60f 100755 --- a/git-clone.sh +++ b/git-clone.sh @@ -72,6 +72,17 @@ Perhaps git-update-server-info needs to be run there?" rm -fr "$clone_tmp" http_fetch "$1/HEAD" "$GIT_DIR/REMOTE_HEAD" || rm -f "$GIT_DIR/REMOTE_HEAD" + if test -f "$GIT_DIR/REMOTE_HEAD"; then + head_sha1=`cat "$GIT_DIR/REMOTE_HEAD"` + case "$head_sha1" in + 'ref: refs/'*) + ;; + *) + git-http-fetch $v -a "$head_sha1" "$1" || + rm -f "$GIT_DIR/REMOTE_HEAD" + ;; + esac + fi } quiet= -- 1.5.2.2.585.g9cc0-dirty ``` ## Alex Riesen, 2007-06-30 13:33 Subject: Re: [PATCH] git-clone: fetch possibly detached HEAD over dumb http Message-ID: <20070630133310.GB2866@steel.home> URL: https://gitlist.dev/e/20070630133310.GB2866%40steel.home In-Reply-To: <20070629083108.GA14747@liacs.nl> ``` Sven Verdoolaege, Fri, Jun 29, 2007 10:31:08 +0200: > + head_sha1=`cat "$GIT_DIR/REMOTE_HEAD"` > + case "$head_sha1" in > + 'ref: refs/'*) > + ;; And what do you do if the HEAD is a reflink on something not in refs/? Like "ref: tmp"? Yes, it is unlikely, but is not forbidden. How about "[0-9a-f]*)" instead: case "$head_sha1" in [0-9a-f]*) git-http-fetch $v -a "$head_sha1" "$1" || rm -f "$GIT_DIR/REMOTE_HEAD" ;; esac ``` ## Sven Verdoolaege, 2007-06-30 13:45 Subject: Re: [PATCH] git-clone: fetch possibly detached HEAD over dumb http Message-ID: <20070630134542.GF7969MdfPADPa@greensroom.kotnet.org> URL: https://gitlist.dev/e/20070630134542.GF7969MdfPADPa%40greensroom.kotnet.org In-Reply-To: <20070630133310.GB2866@steel.home> ``` On Sat, Jun 30, 2007 at 03:33:10PM +0200, Alex Riesen wrote: > Sven Verdoolaege, Fri, Jun 29, 2007 10:31:08 +0200: > > + head_sha1=`cat "$GIT_DIR/REMOTE_HEAD"` > > + case "$head_sha1" in > > + 'ref: refs/'*) > > + ;; > > And what do you do if the HEAD is a reflink on something not in refs/? > Like "ref: tmp"? Yes, it is unlikely, but is not forbidden. It may not be forbidden, but I don't think it would work with current git-clone either. skimo ``` ## Alex Riesen, 2007-06-30 22:23 Subject: Re: [PATCH] git-clone: fetch possibly detached HEAD over dumb http Message-ID: <20070630222336.GA2703@steel.home> URL: https://gitlist.dev/e/20070630222336.GA2703%40steel.home In-Reply-To: <20070630134542.GF7969MdfPADPa@greensroom.kotnet.org> ``` Sven Verdoolaege, Sat, Jun 30, 2007 15:45:44 +0200: > On Sat, Jun 30, 2007 at 03:33:10PM +0200, Alex Riesen wrote: > > Sven Verdoolaege, Fri, Jun 29, 2007 10:31:08 +0200: > > > + head_sha1=`cat "$GIT_DIR/REMOTE_HEAD"` > > > + case "$head_sha1" in > > > + 'ref: refs/'*) > > > + ;; > > > > And what do you do if the HEAD is a reflink on something not in refs/? > > Like "ref: tmp"? Yes, it is unlikely, but is not forbidden. > > It may not be forbidden, but I don't think it would > work with current git-clone either. > Every command which needs a proper .git will not work, so I take this back completely. The check for .git validity includes checking if HEAD contains something sane, and this check is very simple: the HEAD is read (readlink(2) or plain read(2)) and tested if it contains a reference starting with "refs/", which maybe inconsistent with resolve_gitlink_ref, but probably ok. ``` ## Junio C Hamano, 2007-07-01 02:22 Subject: Re: [PATCH] git-clone: fetch possibly detached HEAD over dumb http Message-ID: <7vy7i0amnn.fsf@assigned-by-dhcp.cox.net> URL: https://gitlist.dev/e/7vy7i0amnn.fsf%40assigned-by-dhcp.cox.net In-Reply-To: <20070630222336.GA2703@steel.home> ``` Alex Riesen writes: > The check for .git validity includes checking if HEAD contains > something sane, and this check is very simple: the HEAD is read > (readlink(2) or plain read(2)) and tested if it contains a > reference starting with "refs/", which maybe inconsistent with > resolve_gitlink_ref, but probably ok. Ah, I was not paying close attention to resolve_gitlink_ref(); if it does not require HEAD to point at refs/ I would say it is a bug. Come to think of it, I would further say that we probably should tighten it up a bit: HEAD must be either a valid commit object name (i.e. detached) or a ref that point at somewhere under refs/heads hierarchy, not just anywhere in refs/. ``` ## Alex Riesen, 2007-07-01 16:40 Subject: Re: [PATCH] git-clone: fetch possibly detached HEAD over dumb http Message-ID: <20070701164016.GA3661@steel.home> URL: https://gitlist.dev/e/20070701164016.GA3661%40steel.home In-Reply-To: <7vy7i0amnn.fsf@assigned-by-dhcp.cox.net> ``` Junio C Hamano, Sun, Jul 01, 2007 04:22:04 +0200: > Alex Riesen writes: > > > The check for .git validity includes checking if HEAD contains > > something sane, and this check is very simple: the HEAD is read > > (readlink(2) or plain read(2)) and tested if it contains a > > reference starting with "refs/", which maybe inconsistent with > > resolve_gitlink_ref, but probably ok. > > Ah, I was not paying close attention to resolve_gitlink_ref(); > if it does not require HEAD to point at refs/ I would say it is > a bug. yes, thats why I think its ok. > Come to think of it, I would further say that we probably should > tighten it up a bit: HEAD must be either a valid commit object > name (i.e. detached) That (HEAD must point to a _valid_ commit) will make accidentally corrupted repositories harder to fix. The tool which require a valid repository (cat-file, update-ref, read-tree) are the same tools which you need to fix small problems which can happen, like the commit pointed by HEAD is accidentally pruned from parent repo. ```