IRCloggy #git 2019-08-30

Logs Search ←Prev date Next date→ Channels Documentation

Provider of IRC logs since 2005.
WARNING: As Freenode became unjoinable and lost all warnings in topics, we cannot log channels on Freenode anymore.

2019-08-30

surfist left00:00
surfist joined00:00
qpdbGuest1534100:05
Guest15341 left00:05
omnireq joined00:05
qpdb joined00:05
durham joined00:09
fphilipe_ left00:10
nerosnm joined00:12
Fernando-Basso left00:13
z1haze left00:13
durham left00:14
def_jameblip00:16
nerosnm left00:18
libertyprime left00:20
doug16k joined00:22
learningc left00:25
fatalhalt joined00:30
learningc joined00:36
doug16k left00:40
fphilipe_ joined00:46
g00s joined00:50
liberiga joined00:54
davidfetter_work left00:56
rafalcpp_ joined00:56
rafalcpp left00:57
Oatmeal left01:02
doug16k joined01:05
magic_ninja_work joined01:08
F0rTex left01:12
F0rTex joined01:14
liberiga left01:16
welovfree joined01:16
welovfree left01:16
nerosnm joined01:20
fphilipe_ left01:21
nerosnm left01:24
AtumT left01:29
supernovah joined01:30
supernovah this channel consistently drops me more than others, it's really weird01:30
rafasc the algorithm doesn't like you01:32
Oatmeal joined01:42
nerosnm joined01:42
nerosnm left01:47
fphilipe_ joined01:49
LearnAllTheTime joined01:52
ferdna left01:54
ybyourmom joined01:57
ybyourmom Hey, when I submit a patchset to this particular project, they squash all my commits down before merging it01:57
How do I rebase on top of their squashed commit inside of my local working branch?01:58
I'd like to know if there's a clean way to rebase on top of master after they squashed a bunch of my commits down01:58
rafasc ybyourmom: you mean your branch have some more commits, after you gave them the patch?02:01
Sasazuka left02:01
rafasc and you want to rebase said commits onto their branch?02:02
ybyourmom No, I have say, 5 commits in addition to master -- I submitted a pull request, and they accepted it, and then they merged my 5 commits into master02:02
But when they merged my 5 commits, they squashed them down into 1 single commit02:02
Now I want to merge/rebase the new master into my dev branch02:03
But git can't automatically deduce that my 5 commits are now incorprorated into master02:03
Because the upstream developers went and squashed the 5 commits into 102:03
rafasc why do you say git can't deduce that?02:04
ybyourmom Because in practice when I try to `git pull --rebase upstream master`, git detects the content conflicts as merge conflicts02:05
Is there a way to do things such that git will deduce it?02:05
rafasc git doesn't deduce anything, that's the problem.02:06
ybyourmom git will filter out commits in the master branch that match those in your dev branch02:06
If you *don't* squash the commits down in the master branch02:06
atrigent joined02:07
ybyourmom the problem is that because these people squash the commits down, git can't detect that that 1 squashed commit is actually those 5 commits in my dev branch, which it would usually have just filtered out02:07
I'm asking02:07
rafasc ybyourmom: I think you have the wrong mental model for git.02:07
ybyourmom I'm fairly confident that you're right02:08
Thank god I'm in the #git channel with you though -- I guess that's why I came in here asking02:08
rafasc I didn't mean that as an offence. Sorry if it sounded like that.02:09
ybyourmom It's np02:09
I guess the question is, do you think you understand the scenario I'm describing?02:09
Normally, for the 3 years I've been working with git, it's been filtering out the commits that overlap between my dev branch and the upstream master02:09
Now I'm dealing with a new project where they squash, and it's hampering me02:10
rafasc can you post your !situation?02:10
gitinfo Please post the url returned by `git log -n 20 --all --graph --format="%h %p %d %s" | curl -F text=@- https://upaste.de/` to give us an idea about what your situation is (increase the -n argument if your problem is more complex than 20 commits can explain)02:10
vishal ybyourmom: did you read the 'recovering from upstream rebase' section in the man page?02:10
ybyourmom vishal: no02:11
Is that what I should be reading?02:11
vishal that should be a good starting point02:11
ybyourmom Thanks02:11
rafasc the problem would be easier to explain if you provided the actual example you're at.02:11
ybyourmom nods, secd02:12
ybyourmom https://upaste.de/Zo402:13
rafasc ok, so they squash merged your "exasock-packetdrill-tests" into master.02:14
ybyourmom yes02:15
And my attempts to `git pull --rebase` fail02:15
rafasc on master?02:15
ybyourmom No, my current branch in my local git here is my own dev branch02:16
I'd like to know if the only way to recover here is to `git reset --hard` and then `git cherry-pick`02:16
Or, maybe I should squash my commits on my own branch to match their squash02:16
kerframil left02:17
ybyourmom (And then execute the `git pull --rebase`)02:18
vishal did you try the --onto option to rebase described in that section? that /should/ be able to recover from this, barring any conflicts from any changes they may have made during the ssquash02:18
rafasc vishal: what I am trying to understand, (and failing), is where the --onto should be.02:19
ybyourmom vishal: no, I haven't finished reading it yet02:19
I'm still trying to see if there's a canned answer somebody can give m02:19
rafasc unfortunately, the dev branch isn't in the paste.02:19
ybyourmom I'm perfectly fine with giving you more information in the paste; if you tell me how, I'll do it02:20
vishal --onto master-with-squashed-changes dev~n where n is the number of commits you're transplanting02:21
rafasc vishal: but that may replay commits that were already squashed.02:22
libertyprime joined02:22
vishal is there not a clean 'break' point on your dev branch for the stuff that was merged in vs that's pending?02:23
rafasc I mean, depending on how you choose ~n02:23
vishal I'd imagine not choose the things that went in :)02:23
ybyourmom vishal: I can manage my workflow to make sure that there is always a clear dividing point02:24
rafasc git log --graph --online --bounday dev...origin/master02:24
vishal even if tehre is not a clean dividing point, just rebasse -i your branch to drop the merged commits selectively.. but that becomes a bit more manual than is ideal..02:25
ybyourmom Pretty much02:25
fphilipe_ left02:25
ybyourmom I'm feeling partial to just manually rebasing my dev branch to manually squash the exact same commits they did, and then just doing to normal `git pull --rebase` after that02:26
It'll give me the smooth behaviour I want with less work02:26
I'll read vishal 's recommendations and decide if his thing is better02:26
vishal you won't have the same hashes as upstream, so you'll probably still run into the same problem02:26
rafasc ybyourmom: would you mind pasting the second command I gave you?02:29
ybyourmom Sure, sorry02:29
boombatower left02:29
rafasc no problem. I am just curious if your situation is easily solvable with a rebase --onto02:29
ybyourmom rafasc: fatal: unrecognized argument: --online02:30
rafasc --oneline02:30
sorry02:30
ybyourmom Ah02:30
rafasc I keep making that mistake.02:31
ybyourmom https://upaste.de/hBx02:31
Happens to the biggest dicked among us02:31
bdonnahue left02:32
bdonnahue joined02:34
malina joined02:35
jab416171 left02:36
rafasc ybyourmom: so, the divergence is pretty small... was 5d200c4, was the squash of two commits right? 4d63b41 and e4b532702:36
and you have 91df33f still on your branch, that wasn't included. (going from origin/kofi/exasock-packetdrill-tests)02:38
if that's the case, git rebase --onto origin/master exasock-packetdrill-tests~1 exasock-packetdrill-tests02:39
since it's a simple divergence, there are multiple ways you could achieve the same thing. git reset --hard origin/master; git cherry-pick 91df33f; would be another way.02:40
craftyguy what might cause fetch to hang on rev-list (seen by setting GIT_TRACE) when fetching a remote over ssh? it happens on any remote/repo I try to fetch over ssh. fetching over http works..02:41
vishal rev-list is probably jsut the last thing it does, but isn't hanging on it - I'd look harder on the ssh vs http side of things02:42
rafasc ybyourmom: the problem with pull --rebase, is that it will try to apply 4d63b41,e4b5327,91df33f individually. But since you know the first two are already squashed you can skip those.02:42
vishal e.g. open ports, correct paths/creds etc02:42
craftyguy vishal: oh, maybe trying to ssh manually to the remote or something?02:42
rafasc craftyguy: also try GIT_TRACE2=102:43
vishal sure!02:43
craftyguy GIT_TRACE2 shows a bit more info: https://bpaste.net/show/eBOQ02:44
I can ssh to [email@hidden.address] (the remote in this case). I have other repos with different remotes (e.g. sr.ht, a private git I run, etc) that all show the same issue though.. ssh to them works, but git indefinitely fails to fetch over ssh02:45
rafalcpp_ left02:45
rafalcpp_ joined02:46
craftyguy the failure mode is that it hangs indefinitely, is what I should say02:46
rafasc but did that hang?02:47
craftyguy setting GIT_TRACE2=1 and calling 'git fetch' hung. is that what you are asking?02:48
rafasc yes, there's also GIT_TRACE_PACKET=102:48
craftyguy (it's still hung, 5 minutes later)02:48
rafasc but not sure how that would help02:49
nerosnm joined02:50
craftyguy definitely a lot more stuff printed to terminal, but it still hangs02:51
I see refs in the remote printed to terminal, that is02:51
before it hangs02:52
PityDaFool left02:52
rafasc I think what is missing is the pkt-line.c:80 packet: fetch> 000002:52
that's the server telling it finished sending.02:52
AfroThundr|main joined02:53
craftyguy oh, I actually do see that with GIT_TRACE_PACKET, but it stops immediately after that and hangs: 19:49:40.110484 pkt-line.c:80 packet: fetch> 000002:54
causa_sui joined02:54
lagothrixGuest9575702:54
Guest95757 left02:54
lagothrix joined02:54
nerosnm left02:54
craftyguy the last bit of the output: https://bpaste.net/show/yfES02:55
rafasc after that, I think it's kicks the garbage collector.02:55
jinsun left02:55
Cabanossi left02:57
AfroThundr|main left02:58
rafasc craftyguy: right after that, I get child_exit[0] pid:16309 code:0 elapsed:6.038865 Which is the ssh process.02:58
So for some reason, your ssh command is not finishing..02:58
craftyguy rafasc: thanks for helping, I appreciate it02:59
ssh works when connecting to other systems, so I have no clue why it would not work when git is using it :(02:59
rafasc I think I cannot help you much more than that.02:59
craftyguy no problem, I might be a bit closer to the issue in any case :P03:00
rafasc If you said you are seeing the refs, it means it is working..03:00
MACscr left03:00
AfroThundr|main joined03:00
rafasc it's just not working as it should, I guess.03:00
fphilipe_ joined03:01
craftyguy yea, I guess that's what I meant.. it's not behaving03:01
rafasc anyway, if you find the problem let us know.03:01
craftyguy yea, definitely. I know this channel is logged :)03:01
_ADN_ left03:03
cbreak left03:03
_ADN_ joined03:03
cbreak joined03:03
craftyguy seems like the issue is caused by ssh ControlMaster/Path03:06
if those are set, git gets stuck. if they are not set, git over ssh works03:06
I set that up a loooooong time ago, so I'm not yet sure why this is suddenly a problem03:06
vishal interesting - you dont' really need ssh multiplexing for gitlab, do you? or have you just set it globally?03:07
craftyguy vishal: I have it set globally03:07
the problem wasn't just gitlab, it was any remote I tried to fetch from over ssh03:07
Cabanossi joined03:07
craftyguy including one on my local network03:08
vishal seems like gi, for some reason, waits for the connection to terminate03:10
ybyourmom rafasc: thanks, I'm looking into it03:10
craftyguy vishal: yea that was my conclusion as well. I rolled back to git 2.22 and can reproduce it. ssh on my system is 8.0p1 which is almost a month old, and I've definitely fetched over ssh since then. so I have no idea what changed03:12
nerosnm joined03:12
craftyguy actually that openssh version is a lot older than a month, my distro just updating packaging a month ago03:13
vishal just to confirm, can you try running it with GIT_SSH_COMMAND set to ssh 2>/dev/null03:15
welovfree joined03:16
nerosnm left03:17
zefferno joined03:17
bket left03:23
craftyguy vishal: still hangs when the multiplex stuff is enabled03:23
vishal craftyguy: what if you drop ControlPersist down to say 10s03:27
i expect that'll fix it but doesn't really get us any closer to root causing it :)03:27
HoffmanJesus_Christ03:28
Goplat joined03:28
welovfree left03:28
craftyguy vishal: surprisingly, it doesn't fix it. it still hangs03:29
vishal for over 10s? :)03:30
craftyguy even tried 'ControlPersist 1' just to be sure03:30
vishal try 0 just to be really really sure :)03:30
vishal running out of ideas though03:30
craftyguy 'ControlPersist no' works, but it causes it to terminate immediately :P03:31
so multiplexing wouldn't work elsewhere03:31
vishal right.. making it pointless :)03:31
craftyguy no worries, I've been out of ideas for ~30 minutes, so I really appreciate the help03:32
iam730 left03:35
fphilipe_ left03:36
bket joined03:36
iam730 joined03:37
nic-hartley joined03:38
finn0 joined03:39
libertyprime left03:49
apetresc left03:50
apetresc joined03:51
rchavik joined03:52
pie_ left03:52
SkarmoutsosV joined03:55
libertyprime joined03:56
xelxebar left04:01
xelxebar joined04:02
finn0 Can I adjust the length of a hunk when adding hunk of modified contents in working tree using command `git add -p a.txt`? because one of the hunk covers two different change which I want to commit in seperate commits. I learnt this method of separating changes in different commits in git-stash(1) "#Testing partial commits" example.04:04
patrick042 joined04:07
fphilipe_ joined04:09
paddy left04:10
bn_work left04:11
fjellfras joined04:16
malina left04:18
rafasc finn0: you can split (s), and if that isn't enough, you can edit(e) the patch itself.04:22
ybyourmom later alligators04:23
ybyourmom left04:23
dartmouthed joined04:27
veegee joined04:32
m0viefreak left04:33
fatalhalt left04:38
nerosnm joined04:43
fphilipe_ left04:46
kapilp_ joined04:47
nerosnm left04:47
tsdh joined04:47
magic_ninja left04:49
finn0 rafasc: How? I'm new to git. Could you point me to the documentation or any article related to this?04:49
magic_ninja_work left04:50
rafasc I don't know of any articles on this.04:51
finn0: but for git add -p, it should be simple.04:51
when the hunk shows, just press s.04:51
it will split it into smaller chunks you can accept.04:52
SkarmoutsosV left04:52
rafasc with e) you have to follow the "rules" that are described on the comment.04:52
basically, if you want to ignore a + line, you delete it.04:53
if you want to ignore a - line, you delete the -, and put a space in its place.04:53
(in vim , r<space>) is a good way to do it.04:53
AfroThundr|main left04:54
rafasc then those changes get added to the index, you can commit them, and do another add -p pass.04:54
rinse and repeat as much as needed.04:54
AfroThundr|main joined04:54
magic_ninja_work joined04:58
finn0 don't need any article. git-add(1) man page covers it in detail. Thanks for detailed explaination :)05:01
pks_ joined05:02
pks left05:03
pks_pks05:03
rafasc finn0: also, note that reset/restore/checkout/stash/add all have -p, but the way you edit patches are slightly different.05:05
but there's always a note in the comments explaining.05:06
fphilipe_ joined05:15
rainmanjam joined05:20
MoziM left05:23
rafasc left05:27
nerosnm joined05:28
rainmanjam left05:28
fructose left05:30
Milos_Milos05:31
nerosnm left05:32
fructose joined05:33
Silenced joined05:33
Silenced left05:38
atrigent left05:39
stutter left05:43
abbiya joined05:43
rchavik left05:47
foul_owl left05:52
magic_ninja_work left05:53
MoziM joined05:57
tufteddeer joined05:57
magic_ninja_work joined05:59
atrigent joined06:00
chele joined06:00
tufteddeer left06:02
lalitmee joined06:05
tufteddeer joined06:08
pionen joined06:09
foul_owl joined06:09
pionen left06:09
Anthaas left06:11
Anthaas joined06:13
sQVe joined06:13
nic-hartley left06:15
Noti joined06:20
fjellfras left06:24
Raimondi left06:26
fjellfras joined06:27
g2anj left06:29
SteffanW joined06:30
Raimondi joined06:30
Noti left06:31
iNs left06:35
zefferno left06:35
iNs joined06:35
finn0 left06:35
rafalcpp joined06:35
rafalcpp_ left06:36
rafalcpp left06:36
Raimondi left06:37
Mattiaslndstrm joined06:39
g00s left06:40
Raimondi joined06:40
Mattiaslndstrm left06:42
Oatmeal left06:43
MACscr joined06:43
Mattiaslndstrm joined06:43
lalitmee left06:45
Noti joined06:46
SteffanW left06:47
magic_ninja_work left06:47
khisanth_ left06:55
lankanmon left06:57
lmln joined06:58
oxymoron93 joined06:59
nerosnm joined07:00
mrCyborg left07:02
finn0 joined07:02
mrCyborg joined07:03
nerosnm left07:04
Jackneill joined07:05
ppang joined07:06
Phylock joined07:08
z|bandito left07:12
khisanth_ joined07:13
SteffanW joined07:15
Noti left07:15
yonder left07:19
Silenced joined07:21
Silenced left07:24
fphilipe joined07:25
T_UNIX joined07:28
fphilipe_ left07:28
pyfgcr joined07:31
g2anj joined07:31
igemnace left07:32
mooasaurus left07:35
mooasaurus joined07:36
TomyWork joined07:36
Noti joined07:38
SteffanW left07:38
lalitmee joined07:41
fphilipe left07:49
MoziM left07:49
clime joined07:50
lalitmee left07:50
MoziM joined07:51
lalitmee joined07:54
feriman joined07:55
cd joined07:55
cd left07:56
feriman left07:57
feriman joined07:57
dege joined08:01
fphilipe joined08:02
hussam left08:04
lalitmee left08:05
lalitmee joined08:05
rople joined08:05
rople left08:07
alystair left08:11
hussam joined08:13
lalitmee left08:14
pyfgcr left08:15
lalitmee joined08:16
xcm left08:16
xcm joined08:16
Joel left08:19
floppydh joined08:20
sinhue left08:20
sinhue joined08:20
Viking667 joined08:21
Viking667 really quick question. Has anyone tried installing and running git under wine?08:22
I've had a not quite so good response.08:22
avu why would anyone want to do that though? :)08:23
Viking667 I have a program that runs under Windows, that calls git for one of the things it does. The program doesn't have a Linux equivalent.08:24
(as in, Dolphin Smalltalk doesn't run natively on Linux)08:24
Joel joined08:27
mimajof joined08:28
nerosnm joined08:28
jakobvarmose left08:30
cd joined08:31
jakobvarmose joined08:32
nerosnm left08:33
agowa338 joined08:35
figuinhas joined08:37
Anthaas left08:40
Anthaas joined08:41
mika left08:41
finn0 left08:46
pks_ joined08:51
pks left08:53
pks_pks08:53
ephemer0l_ left08:59
pks left09:04
pks joined09:06
mika joined09:13
Goplat left09:23
ave_milia joined09:25
[twisti] joined09:26
[twisti] can i somehow make a .gitattributes rule that covers files with specifically NO file ending ? i dont want the rule to just cover all files, just files with no ending09:26
pks left09:28
zamba left09:34
tombusby left09:34
pyfgcr joined09:35
tombusby joined09:35
howdoi joined09:36
ephemer0l_ joined09:36
_ikke_ [twisti]: no, that does not exist09:38
learningc left09:39
[twisti] bleh. thank you. at least i can stop searching then09:40
patrick042paddy09:41
mika left09:42
zamba joined09:42
[twisti] left09:47
igemnace joined09:57
nerosnm joined09:58
nerosnm left10:03
pyfgcr left10:07
thijso left10:08
f8e3 is it Docs.md or DOCS.md like readme too?10:10
GyrosGeier that's just a convention10:12
the README being written uppercase is a convention from the eighties10:13
because back then, ls sorted files by ASCII value, and you'd use lowercase characters for all other files, so README sorted first10:13
also, it stands out a bit10:14
nerosnm joined10:14
paws_ joined10:16
f8e3 aye ok10:16
fjellfras left10:18
learningc joined10:19
Phylock left10:19
ilogger2 joined10:24
redrabbit joined10:24
uidzer0 joined10:24
karoshii joined10:24
bin_bash joined10:24
PacKetSlayer joined10:24
timthowtdi joined10:25
SirFunk joined10:25
kegster joined10:25
d^sh joined10:25
uplime^ joined10:25
waveclaw joined10:26
qgTG joined10:26
Lunatrius joined10:26
raatiniemi joined10:26
MACscr joined10:26
squirrel joined10:26
scivola joined10:26
Defcronyke joined10:27
yuljk joined10:27
Mattiaslndstrm joined10:27
mowcat joined10:27
sborza joined10:27
bleb joined10:27
tkazi joined10:27
NightMonkey joined10:27
rigid joined10:27
dansan joined10:27
mescalinum joined10:27
zerogue joined10:27
alip joined10:27
yourfate joined10:27
Milos joined10:27
donofrio joined10:27
WSDMatty joined10:27
flying_sausages joined10:27
Ram-Z joined10:27
mu joined10:27
wolfshappen joined10:27
Karyon joined10:27
raven-au joined10:28
dgw joined10:28
cdown joined10:28
ldh joined10:28
ChrisLane joined10:28
webstrand joined10:28
Noti joined10:28
donofrio left10:28
donofrio joined10:28
immae joined10:28
n000g joined10:29
nkuttler joined10:29
j7k6 joined10:29
remedan joined10:29
afoo joined10:29
ikonia joined10:29
sunri5e joined10:29
wyoung joined10:29
deltab joined10:29
Rhonda joined10:29
lavalike joined10:29
justanotheruser joined10:29
nasamuffin joined10:29
brethil joined10:29
blaisebool joined10:29
mooasaurus joined10:29
g2anj joined10:30
royal_screwup21 joined10:30
unreal joined10:31
rkta joined10:31
SCHAPiE joined10:31
Mattiaslndstrm left10:32
kjartan joined10:32
Mattiaslndstrm joined10:33
Phylock joined10:33
chele joined10:33
Swant joined10:33
kini joined10:33
kgz joined10:34
bipolar joined10:34
cim209 joined10:34
finn0 joined10:35
SteffanW joined10:36
Ekho joined10:36
Noti left10:37
pinux joined10:39
apetresc joined10:40
jabular joined10:40
tribly joined10:40
doug16k joined10:41
Chaser joined10:41
foul_owl joined10:41
learningc joined10:41
pinux left10:42
Tobbi_ joined10:45
mika joined10:48
bolovanos joined10:48
jaeckel joined10:49
bolovanos left10:49
bolovanos joined10:50
pks joined10:52
nerosnm joined10:53
f8e3 git remote add foo <can use hostname istead of ip here?>10:55
crose joined10:55
f8e3 its desktop on the local wifi10:56
but dynamic adresses change thats why i ask10:56
GyrosGeier absolutely10:57
nerosnm left10:57
f8e3 i have two computers with same hostname 'debian' on the local lan?10:59
can i use MAC, (i dont want static ip in route rmenue although possible)10:59
GyrosGeier no, MAC addresses are not visible to git11:01
your zeroconf setup should try to give you stable hostnames11:01
also, I'd rename at least one of the machines11:02
f8e3 changed static hostname with hostnamectl, new add to repo: ssh: Could not resolve hostname debian-mbox: Name or service not known11:04
hostA cant push since hostB (whose name changed not hostname-findable)11:05
GyrosGeier do you have a zeroconf/mdns setup where you can resolve names like "debian.local"?11:05
the .local might be mandatory unless it's in your search path11:06
Viking667 f8e3: how the heck did you manage that little trick?11:06
Normally if two hosts have the same name, there's trouble in the DNS server.11:06
emsjessec joined11:07
GyrosGeier well, the DNS server needs to be told first11:08
Viking667 f8e3: First, can you ping its ip? Then, can you ping it by hostname?11:08
GyrosGeier there are so many historic ways to get hostname lookup on a LAN11:08
Viking667 Agreed. Those are the two I use to begin with, they have the advantage of working on any OS.11:09
parsnip joined11:10
f8e3 no trick just install deb hostname=debian; i told the router to chagne hostname of the pingable ip, now it is ok11:10
Viking667 And doesn't hostnamectl only change /etc/hostname ?11:11
f8e3: cool.11:11
Anyhow, I'm out of here.11:11
bye all11:11
f8e3 thank yall11:11
lalitmee joined11:11
g2anj left11:12
mowcat left11:12
luis joined11:16
nerosnm joined11:17
gitinfo set mode: +v11:17
g2anj joined11:21
crose left11:22
f8e3 On Permission denied (publickey). do i need to generete new ssh keys and copy the pubkey to the target push host?11:22
or can i use whats in ~/.ssh *.pub ?11:22
SteffanW left11:22
SteffanW joined11:23
jimender2 joined11:24
nerosnm left11:25
jottr_ joined11:26
rafasc joined11:28
learningc left11:31
mischief joined11:33
mns joined11:34
lalitmee left11:36
Mattiaslndstrm left11:37
tens0r joined11:37
Silenced joined11:38
GyrosGeier f8e3, you can use what's there11:40
ssh-copy-id(1) is useful too11:40
lalitmee joined11:45
tens0r left11:46
Noti joined11:47
SteffanW left11:47
nerosnm joined11:53
tens0r joined11:57
tens0r left11:57
finn0 left12:01
tens0r joined12:02
freeman42x[NixOS joined12:04
royal_screwup21 left12:06
tens0rts0r12:07
SkarmoutsosV joined12:08
ts0rtens0r_12:09
tens0r_tens0r12:10
tens0r left12:10
tens0r joined12:11
lalitmee left12:16
AtumT joined12:16
lalitmee joined12:16
m0viefreak joined12:17
lalitmee left12:18
lalitmee joined12:18
SteffanW joined12:19
Noti left12:20
ExeciN joined12:21
tens0rtx0r12:22
malina joined12:26
Noti joined12:28
SteffanW left12:28
freeman42x[NixOS left12:32
qotta joined12:33
thijso joined12:35
fahadash joined12:38
Peng joined12:38
jomo_ joined12:40
st-gourichon-fid joined12:40
tx0r left12:41
freeman42x[NixOS joined12:44
tens0r joined12:45
tens0r left12:46
nerosnm left12:48
Eryn_1983_FL joined12:49
crose joined12:51
jottr_ left12:56
lalitmee left13:03
thijso left13:03
octav1a joined13:04
thijso joined13:06
CodeSlingerPaul joined13:07
cdown_ joined13:09
cdown left13:11
cdown_cdown13:12
RhineDevil joined13:13
barteks2x joined13:13
RhineDevil How can I check from a script if a http/https repo wants a password or ssh repo key wants a password?13:15
(obviously git repos)13:16
Habbie git <something> --dry-run?13:16
rafasc RhineDevil: what's the context for this?13:17
dummy credential helper?13:17
RhineDevil rafasc, a git wrapper that should download automatically lots of repos13:17
Mattiaslndstrm joined13:18
RhineDevil It would be better if the ones needing a password would be prompted first13:18
rafasc what should happen if the repo requires a password?13:18
RhineDevil rafasc, the repo requiring a password will be prompted first, for the others the user just have to wait13:18
Habbie, tried git clone <repo> --dry-run and git clone --dry-run <repo>, no result13:20
It says it doesn't know an option called dry-run13:21
Habbie i do suppose --dry-run does not make sense for clone13:21
storrgie joined13:21
rafasc I'd say: for https, configure a credential helper, and use something like git ls-remote on all of then.13:21
Inline joined13:22
rafasc this should be quick, and allow you to populate the credential-helper with valid credentials. Then run your clone in a loop.13:22
and it should use the credentials from before.13:22
still doesn't cover for ssh with passphrases.13:23
RhineDevil rafasc, how much is the default credential timeout?13:23
rafasc RhineDevil: you can configure it. I think 300s13:24
thijso left13:24
rafasc actually 900.13:24
but that's just the '-cache', you can use the '-store' to save it permanently.13:25
thijso joined13:25
bremner +-13:25
RhineDevil rafasc, by changing the user's settings or by passing arguments?13:25
rafasc if it fails to authenticate, it gets removed from the store.13:25
I guess that's why bremner said +-13:26
nerosnm joined13:26
rafasc RhineDevil: I think it can be done both ways. git -c credential.helper 'cache --timeout 200' clone url.13:27
omnireq joined13:27
RhineDevil rafasc, so I will cache password instead of git?13:28
rafasc it is still git caching the password.13:29
RhineDevil Why I set the credential helper then?13:29
jottr_ joined13:30
rafasc Because the idea would be to authenticate first on a fast command like ls-remote, and make git use the cached credentials when cloning.13:30
for url in list_of_repos: do git ls-remote url; done; ... then another loop for clone.13:32
just an idea.13:32
nerosnm left13:32
rafasc the -cache helper also has a '--socket <path>', if you want to store the cache on a different place.13:33
monr0e joined13:33
rafasc the cache will be stored in memory, with that I mean if you want to use a different endpoint.13:34
macroprep_ joined13:34
RhineDevil Thanks a lot13:37
rafasc For ssh, I don't know how to solve it without using an ssh-agent13:40
DWestCoast joined13:45
tufteddeer joined13:46
RhineDevil left13:48
freeman42x[NixOS left13:49
freeman42x[NixOS joined13:51
tufteddeer left13:51
tryte joined13:51
tufteddeer joined13:52
john2496 joined13:52
nerosnm joined13:52
malina left13:54
moldybits` joined13:54
malina joined13:55
tufteddeer left13:56
nerosnm left13:56
thijso left13:57
boombatower joined13:57
feriman joined13:58
nerosnm joined13:59
pycer joined13:59
thijso joined13:59
czart joined14:01
pycer left14:01
Es0teric joined14:01
moldybits`moldybits14:07
silverballz joined14:09
silverballz left14:09
silverballz joined14:09
silverballz left14:09
silverballz joined14:09
silverballz left14:13
silverballz joined14:13
astronavt joined14:16
omnireq left14:17
gareppa joined14:19
macroprep__ joined14:19
luckman212 joined14:20
macroprep_ left14:23
freeman42x[NixOS left14:23
NL3limin4t0r joined14:23
NL3limin4t0r left14:24
hendry joined14:24
DWestCoast left14:26
SkarmoutsosV left14:27
jottr_ left14:31
jottr_ joined14:32
chele left14:34
SkarmoutsosV joined14:35
silverballz left14:40
silverballz joined14:41
silverballz left14:41
silverballz joined14:41
mooasaurus left14:42
mooasaurus joined14:42
thijso left14:45
greggerz joined14:45
royal_screwup21 joined14:46
monr0e left14:46
Mattiaslndstrm left14:50
WSDMatty left14:51
WSDMatty joined14:52
webstrand I need to commit the contents of a git repository to a subversion repository.14:53
Is there some way I can flatten the .git folder to a single file that just contains refs and remotes?14:53
fraya joined14:53
rafasc webstrand: that sounds... wrong.14:54
webstrand Basically, I'd like to keep track of where the files came from, but I don't want to store the full repository history in subversion14:55
kinda like a worktree, just referencing a remote rather than a local file14:56
rafasc git bundles can turn a repository into a file.. But I think it doesn't make sense tracking it in a svn repo.14:56
fflam joined14:56
rafasc just add the files, and document it in the commit message?14:57
webstrand That's what I'll do, if there's no better solution.15:00
jottr_ left15:00
rafasc webstrand: is this being used with github?15:01
webstrand Yep, it's a clone of a repository from github15:02
averell joined15:02
rafasc I think github still serves its repos as svn as well. So you could probably add it as an external.15:02
but I have very little experience with that.15:02
webstrand I have very little experience with that too, I recall it being very fragile15:02
Didn't know github ever served svn repos15:03
paddy joined15:03
Noti left15:04
tufteddeer joined15:04
pyfgcr joined15:07
webstrand Maybe I can do something with git unpack-objects and a shallow clone?15:07
rafasc for getting the files? check !deploy15:08
gitinfo Git is not a deployment tool, but you can build one around it (in simple environments) or use it as an object store(for complex ones). Here are some options/ideas to get you started: http://gitolite.com/deploy.html15:08
webstrand My idea was to unpack the pack files, set .git/shallow to a single commit, and then use `git gc` to repack only that commit.15:11
it seems to have worked15:11
Not as much size reduction as I'd hoped, though15:12
binarysamsa joined15:12
malina left15:13
malina joined15:14
bluesmonk joined15:14
rafasc tracking a shallow repo inside a svn repo seems like an insane idea.15:14
malina well 55015:14
and now wiht hexchat 57015:14
so I had some browsers open but say ob is hitting 650 base then15:15
rafasc malina: wrong chat?15:15
malina still high but fair enough15:15
ah yes, sorry :D15:15
binarysamsa left15:16
WSDMatty left15:16
malina left15:16
WSDMatty joined15:17
malina joined15:17
omnireq joined15:18
malina left15:18
karstensrage joined15:19
binarysamsa joined15:21
WSDMatty left15:23
WSDMatty joined15:25
nerosnm left15:30
kamog joined15:33
ToBeFree joined15:34
gitinfo set mode: +v15:34
WSDMatty left15:34
nerosnm joined15:35
bdonnahue joined15:36
binarysamsa left15:37
freeman42x[NixOS joined15:38
bluesmonk hello guys, I think I've messed up my repo. I've created a summary of events here https://bpaste.net/show/eY-M15:39
halp please15:40
john2496 left15:40
bluesmonk it's in markdown, but I don't know of any bpaste site for sharing formatted markdown15:41
Habbie gist.github.com can render markdown - it does require a github account15:41
i cannot immediately help with your problem but i have to say you did very well in documenting it!15:41
bdonnahue left15:42
bluesmonk thanks!15:43
rafasc Habbie: was not aware github deprecated anonymous gists.15:45
Habbie rafasc, i.. checking15:45
rafasc, i don't see how to make one :)15:45
rafasc Habbie: https://github.blog/2018-02-18-deprecation-notice-removing-anonymous-gist-creation/15:46
it was possible before..15:46
bluesmonk https://gist.github.com/diegoquintanav/d130ee3ac88d748f5991d99d8f82c2bd15:47
Habbie rafasc, ah15:47
bluesmonk yeah that's why I wanted to avoid it :D15:47
but anyway15:47
binarysamsa joined15:48
bluesmonk it has a tl;dr15:49
:shiny:15:49
rafasc bluesmonk: you were at the develop branch>15:50
?15:50
bluesmonk yes15:50
rafasc do this: git log --graph --oneline develop@{1};15:51
increase the number, until you see what you want to see.15:51
bluesmonk I've updated the gist with some data I forgot to add.15:52
rafasc I think it is 8b5f239,15:53
bluesmonk rafasc: I'll do that15:53
rafasc the pre-rebase state.15:53
bluesmonk I want to move HEAD of develop to that commit15:53
or I think I want to do that15:53
rafasc git checkout develop; git reset 8b5f23915:54
(possibly --hard)15:54
bluesmonk by reverting my interactive rebase should be perfect15:54
nerosnm left15:54
bluesmonk let's see15:54
only stuff I lose with --hard is: changes not staged, not commited, not stashed, right?15:54
astronavt is there a difference between reset --soft and branch -f15:54
i use the latter when i need to move branch heads around because i can never rememeber all the different reset incantations and im afraid to mess up15:55
rafasc astronavt: I don't think so.15:55
vavkamil joined15:56
rafasc the main difference is that reset operates on the current checked out branch.15:56
while branch -f, allows you to do : git branch -f master 123123; git branch -f develop 456456;15:56
without having to checkout master or develop15:56
nerosnm joined15:57
bluesmonk rafasc: question15:57
rafasc git reset --soft X == git branch -f HEAD X;15:57
bluesmonk: answer15:57
tufteddeer left15:57
bluesmonk git log --graph --oneline develop@{1} > do you suggest increasing the index until I reach commit 8b5f239?15:58
what for?15:58
tufteddeer joined15:58
SkarmoutsosV left15:58
rafasc No, if you know that 8b5f239 looks correct, you can reset to it.15:58
bluesmonk ah okay15:58
Sasazuka joined15:59
bluesmonk what would be the right way to revert the interactive rebasing based on the reflog index?15:59
rafasc you can use git reset --hard develop@{1}15:59
bluesmonk that would reset the last operation15:59
rafasc if developt@{1} resolves to that commit.16:00
bluesmonk I see16:00
ok brb16:00
bluesmonk pats himself on the back16:00
macroprep_ joined16:01
impermanence joined16:02
rafasc you could do: git reflog develop; search for the desired commit, etc.16:02
scheming_around joined16:02
Sasazuka__ joined16:02
rafasc it can be hard to guess the correct number, especially since you performed multiple rebases..16:02
davidfetter_work joined16:02
macroprep__ left16:02
scheming_around Why do tag refs point to "tag objects", which point to the actual commit the tag refers to?16:03
I.e., why do tag objects exist?16:03
rafasc instead of using reflog which defaults to reflog HEAD, reflog develop can sometimes give you less entries. Easier to find the correct commit.16:03
feriman left16:04
bluesmonk another question: each time I raise the index on `git log develop@{1}` I get different commits for the same commit logs, until I get to develop@{3}, which shows commit 8b5f23916:04
rafasc scheming_around: tag objects exist because sometimes you want to store information about who tagged, when it was tagged, etc.16:04
Sasazuka left16:05
bluesmonk rafasc: so should it be safe to do git reset --hard develop@{3}?16:05
rafasc bluesmonk: those are your previous rebase results. The ones you quit, but still ran.16:05
binarysamsa left16:05
oatmealraisin joined16:05
bluesmonk rebase changes my commits then?16:06
john2496 joined16:06
rafasc bluesmonk: next time, if you want to abort, either delete everything, or in vim use :cq16:06
bluesmonk: that will "crash" the editor, and nothing will be done.16:06
malina joined16:06
bluesmonk cool16:06
malina left16:06
malina joined16:06
rafasc bluesmonk: use git log --graph --oneline develop@{3}; if it looks good, reset --hard dev@{3}16:07
webstrand rafasc: Looks like if I delete all of the objects except for the HEAD tree object, the repo still appears to function, but fetch and pull no longer work16:08
rafasc scheming_around: and the tag always refers to the object id, the hash.16:09
bluesmonk yasss16:09
rafasc: thank you sir16:09
it worked16:09
or lady, dunno16:09
person16:09
rafasc scheming_around: so you can have tags point to blobs, other tags, commits, trees etc...16:09
I identify as a bilogical automaton.16:10
bluesmonk I will default to sentient being from now on16:10
rafasc you're welcome, glad I could help.16:10
bluesmonk on the internet no one know you are a dog16:10
thanks again!16:11
rafasc bluesmonk: and btw, ignore that SO post. It's outdated. You should be using --rebase-merges instead.16:11
bluesmonk what are the drawbacks of rebasing merge commits?16:11
if any16:12
rafasc None, really... Some people don't like merge commits.16:12
bluesmonk ah I like them :)16:13
rafasc Mostly because people misusing merges gave it a bad rep.16:13
seaef joined16:13
bluesmonk how's that?16:13
davidfetter_work o/` merges don't give a damn about a bad reputation o/`16:13
fission6 joined16:14
thijso joined16:15
tufteddeer left16:17
seaef left16:17
bluesmonk rafasc: last thing, and for the sake of understanding how develop@{} works... what part of the reflog index would be equivalent to doing develop@{3}?16:18
or is there no equivalence?16:18
scheming_around rafasc I see, ty16:19
rafasc scheming_around: I said object id, as a vague term. tags can point to any object really. blobs/trees/commits/other tags/16:19
and the tag-object is the difference between git 'lightweight' and 'annotated' tags16:20
storrgie left16:21
rafasc @{3} means the third entry on the reflog. master@{3} = 3rd entry of git reflog master;16:21
tufteddeer joined16:22
xcm joined16:23
storrgie joined16:24
ExeciN left16:24
bluesmonk I see, I've never used reflog before with an argument16:24
what's the default argument? --all?16:25
sorry I should be googling that16:25
git reflog == git reflog show HEAD16:26
thiago joined16:26
Silenced left16:29
gareppa left16:30
fraya left16:36
thijso left16:36
paraxial joined16:38
freeman42x[NixOS left16:38
Es0teric left16:38
fission6 left16:39
m1rror joined16:41
paraxial left16:43
paraxial joined16:46
bdonnahue joined16:46
malina left16:47
thijso joined16:48
royal_screwup21 left16:48
magic_ninja_work joined16:53
thijso left16:54
nic-hartley joined16:55
ara joined16:56
paddy left16:58
durham joined16:58
thijso joined17:00
xcm left17:02
xcm joined17:03
durham left17:03
pyfgcr left17:06
thijso left17:12
Kaivo joined17:12
binarysamsa joined17:13
liberiga joined17:20
thijso joined17:20
malina joined17:21
m0viefreak left17:24
binarysamsa left17:27
ara left17:30
nerosnm left17:30
orbyt_ joined17:31
ExeciN joined17:37
oatmealraisin left17:37
dijitol joined17:43
macroprep_ left17:47
lucasb joined17:50
kjartan left17:51
kjartan joined17:54
nic-hartley left17:55
f8e3 how to .gitignore folder foo/ but not file foo/bar.jar ?17:57
uplime^also_uplime18:01
cixx joined18:02
cixx hi.18:02
is there a working tool to convert TFS history to Git ?18:02
f8e3 cixx hi cixx18:03
nic-hartley joined18:04
cixx hi f8e3 hi18:08
jab416171 joined18:10
WSDMatty joined18:13
belgin joined18:13
WSDMatty left18:13
belgin hello18:13
m1dnight1 joined18:14
belgin i have a patch and it has To: specified but send-email still asks me where to send it. why?18:14
m1dnight1 left18:14
WSDMatty joined18:14
magic_ninja_work left18:15
fission6 joined18:29
WSDMatty left18:29
JanC joined18:33
bluesmonk f8e3: add foo, in the next line add !foo/bar.jar18:33
belgin left18:34
bluesmonk sorry, first line should be foo/*18:34
don't know if foo alone does it :P18:35
it should though18:35
alystair joined18:38
f8e3 true18:38
hsiktas joined18:39
fission6 left18:40
bluesmonk left18:40
justache joined18:41
xcm left18:42
oatmealraisin joined18:42
xcm joined18:44
fraya joined18:44
leeN joined18:45
Kaivo I have documentation committed that we also push on a CMS. The documentation may be edited in the CMS, and in the git repo. I need to reconcile the changes between the last tag in the repo, the current CMS versin and the current file in repo (maybe not stagged yet). I have the 3 distinct versions of the doc already. How would I go about creating a not commited merge of those 3 versions, maybe causing a conflict18:51
just like when doing a git merge or git stash pop.18:51
binarysamsa joined18:54
malina left18:55
m1dnight_ joined19:02
binarysamsa left19:03
Kaivo left19:11
Kaivo joined19:12
Kaivo Ok, I figured it out. `git merge-file -p file.remote file.tagged file.current`19:13
m1dnight1 joined19:15
m1dnight_ left19:17
jottr_ joined19:21
SkarmoutsosV joined19:21
assemblyman joined19:22
dijitol left19:23
john2496 left19:23
sbeyer joined19:24
fraya left19:24
mowcat joined19:24
mowcat left19:25
cagomez joined19:25
dijitol joined19:25
mowcat joined19:25
cixx why it is not possible to convert tfs history to git completely?19:28
it gets only part of history on some solutions but not all of them19:28
czart left19:29
theoceaniscool joined19:35
m1dnight1 left19:36
orbyt_ left19:37
m1dnight1 joined19:37
Milos left19:42
assemblyman left19:42
Milos joined19:43
czart joined19:46
czart left19:47
nic-hartley left19:47
freeman42x[NixOS joined19:50
assemblyman joined19:59
emsjessec left20:01
realies joined20:01
thiago left20:01
realies left20:02
paddy joined20:02
orbyt_ joined20:03
level3 joined20:03
ericfreese joined20:05
nerosnm joined20:06
thiago joined20:06
nerosnm left20:10
shicks2525 joined20:10
arecaceae joined20:12
SwiftMatt joined20:27
jottr_ left20:32
shicks2525 left20:33
jcbitter joined20:34
feriman joined20:36
shicks2525 joined20:36
ericfreese Hi all, I'm wanting to exclude symbolic refs from a `git for-each-ref` command. I was kind of looking for a `--no-symbolic` (or similar) flag but it doesn't seem to exist. I've gotten something working by using `%(symref)` in the format string and piping through grep filter the sym refs out, but I'm wondering if there is a better way?20:40
And as a follow-up question if there's not a better way, would it make sense to try to submit a patch that adds `--symbolic` and `--no-symbolic` flags?20:41
Stoot joined20:46
realies joined20:48
assemblyman left20:49
greatgatsby joined20:49
kamog left20:49
wootehfoot joined20:51
liberiga left20:53
lmln joined20:54
polman joined21:02
m1dnight1 left21:06
m1dnight1 joined21:07
SkarmoutsosV left21:09
azwieg103 joined21:10
omnireq left21:17
sbeyer left21:17
orbyt_ left21:18
azwieg103 left21:20
azwieg103 joined21:21
|Fitz| joined21:23
orbyt_ joined21:28
bolovanos left21:28
Guest15733 joined21:28
kerframil joined21:30
Cabanoss- joined21:33
random_yanek joined21:33
sgn_ joined21:36
Atlenohen joined21:39
bdonnahue left21:39
karoshi joined21:40
karoshi See if you have amendments to some strangers code on Git. Switch branch and push? Do they get notified? Can't see it listed anywhere21:41
octav1a left21:42
octav1a joined21:42
kerframil left21:43
kerframil joined21:43
impermanence left21:45
vavkamil left21:46
bdonnahue joined21:48
bdonnahue left21:53
oatmealraisin left21:54
feriman left21:55
diogenese joined21:58
jrnieder joined22:00
cagomez left22:00
wootehfoot left22:01
cagomez joined22:02
greggerz left22:03
infinisil joined22:05
tufteddeer left22:06
tufteddeer joined22:06
silverballz left22:07
kerframil left22:08
freeman42x[NixOS left22:08
Kaivo left22:10
tufteddeer left22:11
kerframil joined22:11
orbyt_ left22:11
tufteddeer joined22:13
infinisil Hey, while trying to use git to produce deterministic results, I came across a problem22:13
Namely that the .git/objects/info/packs file doesn't have a deterministic order of .pack files22:13
random_yanek left22:14
random_yanek joined22:14
infinisil This probably happens during a `git repack` operation that I'm running22:15
nic-hartley joined22:15
infinisil I tried finding the place in git's source where this happens, but I couldn't find it22:15
Can anybody tell me where in git's source this file gets written to?22:16
It contains lines of the form "P pack-f01fe51bef092064a1808d5a5e3da9419fbfcfab.pack" for me22:16
Ah it might be here https://github.com/git/git/blob/master/server-info.c#L33922:19
The name of the file initially made me not consider it22:19
Phylock left22:21
gxt joined22:24
ericfreese left22:25
azwieg103 left22:27
omnireq joined22:27
xcm left22:31
xcm joined22:31
freeman42x[NixOS joined22:31
infinisil Well I have no idea how to make it deterministic :(22:33
The source of non-determinism is most likely file listing order22:34
meLon joined22:34
thiago left22:34
theoceaniscool left22:36
cagomez left22:39
kerframil left22:42
lucasb left22:46
orbyt_ joined22:54
dijitol left22:57
yonder joined23:00
xcm left23:03
xcm joined23:04
infinisil Okay so I think I found a way of preventing this determinism23:04
git gc --keep-largest-pack23:04
"All packs except the largest pack and those marked with a .keep files are consolidated into a single pack. When this option is used, gc.bigPackThreshold is ignored."23:04
This seems like it would always make it a single pack instead of multiple23:05
Atlenohen left23:05
CodeSlingerPaul left23:07
magic_ninja joined23:12
jrnieder infinisil: do you mean you want byte-for-byte reproducibility?23:15
Sleepy63 joined23:15
infinisil Pretty much yeah23:15
tufteddeer left23:15
tufteddeer joined23:16
Wizek joined23:16
cagomez joined23:17
tufteddeer left23:20
leeN left23:25
m1dnight1 left23:26
m1dnight1 joined23:27
jrnieder any more context? why? :)23:27
(something like !xy, but not quite)23:28
gitinfo You are asking about a solution, but we can help you a lot better if you could describe the actual problem you are trying to solve.23:28
infinisil I'm packaging software for Nix, which only allows network access for deterministic processes (by forcing you to provide a sha256 hash upfront)23:29
jrnieder oh man23:29
thanks, that's interesting. You can use "git fast-export" output as the deterministic thing23:29
instead of the git repo itself23:30
dzho yeah git is probably a thing very much not to use in something that's meant to go into nix23:30
infinisil In most cases this isn't a problem, because we can just get rid of the whole .git directory23:30
But sometimes packages do fancy things with .git to build23:30
moldybits left23:30
jrnieder I'm not sure fast-export's output order is deterministic, but it would be reasonable to make it so23:30
infinisil Huh interesting, never seen fast-export before23:31
moldybits joined23:33
LearnAllTheTime joined23:34
infinisil Well for now we use this procedure to make it as deterministic as possible: https://github.com/NixOS/nixpkgs/blob/00762b57545dba3b3e21e3a649b574db3db36c65/pkgs/build-support/fetchgit/nix-prefetch-git#L235-L27023:34
And while this isn't truly deterministic, it's worked well enough23:35
tufteddeer joined23:35
jrnieder is very afraid23:35
infinisil Except that I just came accross this problem of pack file ordering not being the same23:35
Yeah it doesn't look that good eh hehe23:35
m1dnight1 left23:36
infinisil jrnieder: What makes you think that fast-export would be good for that?23:36
jrnieder it's a standardized format, better for archival than the .git dir itself. There's no threading-induced nondeterminism23:37
m1dnight1 joined23:37
infinisil I see23:37
Guest15733 left23:37
thiago joined23:38
infinisil I'll maybe look into switching nixpkgs to use something like this instead23:38
For now this --keep-largest-pack fix seems to be the easiest workaround23:39
jrnieder if you're just worried about .git/objects/info/packs, you can "rm" it23:39
infinisil Good point, isn't that needed for anything though?23:39
jrnieder not unless you're serving over raw ("dumb") HTTP23:40
infinisil I see, that might be better then23:40
m0viefreak joined23:44
infinisil Although, with --keep-largest-pack, this won't influence packages which only have a single pack file already23:48
freeman42x[NixOS left23:48
infinisil So I guess I'll go with that after all23:48
Means less work updating sha256's23:48
crose left23:52
freeman42x[NixOS joined23:52
greatgatsby left23:56

Logs Search ←Prev date Next date→ Channels Documentation