Problem with GitHub repos ?

[Francois-Xavier Le Bail]

Hi,

are there update problems with the tcpdump & libpcap git repos ?


Many new branches and bad Travis runs.

See the attachments ...


Greetings,
Francois-Xavier

[Guy Harris]

Post by Francois-Xavier Le Bail
are there update problems with the tcpdump & libpcap git repos ?
Many new branches and bad Travis runs.

Which new branches, and...

Post by Francois-Xavier Le Bail
See the attachments ...

...which attachments? (I don't know whether tcpdump-workers always rejects attachments or only sometimes rejects them, but people have had problems with them.)

If the Travis errors that showed up a while ago in my mailbox are examples of the bad Travis runs, then the branches to which they refer appear to reflect branches in the bpf.tcpdump.org repository, at least for tcpdump. Those repositories get pushed, in some fashion, to the GitHub repository (Nightly? Periodically with some other period? Automatically when changes get pushed to the bpf.tcpdump.org repository? Other?); if that results in the GitHub repositories having the same branches as the bpf.tcpdump.org repositories, that might explain the many new branches.

[Francois-Xavier Le Bail]

Post by Guy Harris

  are there update problems with the tcpdump & libpcap git repos ?
  Many new branches and bad Travis runs.

Which new branches, and...

  See the attachments ...

...which attachments?  (I don't know whether tcpdump-workers always rejects
attachments or only sometimes rejects them, but people have had problems with
them.)

They are here : https://gist.github.com/fxlb/10525029

Which new branches,

Post by Guy Harris
If the Travis errors that showed up a while ago in my mailbox are examples of
the bad Travis runs, then the branches to which they refer appear to reflect
branches in the bpf.tcpdump.org repository, at least for tcpdump.  Those
repositories get pushed, in some fashion, to the GitHub repository (Nightly? 
Periodically with some other period?  Automatically when changes get pushed to
the bpf.tcpdump.org repository?  Other?); if that results in the GitHub
repositories having the same branches as the bpf.tcpdump.org repositories, that
might explain the many new branches.

Just suprising to see e.g. 4 branches -> 21 branches in one shot...
Thanks for explaining, I see now.

[Michael Richardson]

Post by Guy Harris

Post by Francois-Xavier Le Bail
are there update problems with the tcpdump & libpcap git repos ?
Many new branches and bad Travis runs.

Which new branches, and...

many branches with FOO_4.5, and some new ones with FOO_4_5.

Post by Guy Harris
If the Travis errors that showed up a while ago in my mailbox are
examples of the bad Travis runs, then the branches to which they refer
appear to reflect branches in the bpf.tcpdump.org repository, at least
for tcpdump. Those repositories get pushed, in some fashion, to the
GitHub repository (Nightly? Periodically with some other period?

They get pushed nightly, but I added --all so that all the old branches would
get pushed. Should be a one-time event.

--
] Never tell me the odds! | ipv6 mesh networks [
] Michael Richardson, Sandelman Software Works | network architect [
] ***@sandelman.ca http://www.sandelman.ca/ | ruby on rails [

[Francois-Xavier Le Bail]

Post by Francois-Xavier Le Bail
Hi,
are there update problems with the tcpdump & libpcap git repos ?
Many new branches and bad Travis runs.
See the attachments ...

the images are here :
https://gist.github.com/fxlb/10525029

[Guy Harris]

Post by Francois-Xavier Le Bail
https://gist.github.com/fxlb/10525029

Yes, those branches are branches that exist in the bpf.tcpdump.org repositories; I guess pushing from there to GitHub causes the GitHub repository to have them as well.

Most of them are not worthy of being run through Travis, at least as far as I'm concerned; can it be configured to build only the trunk (master) for now? (I think the next releases will be 1.6/4.6, coming from the trunk; once those are branched, the 1.6 and 4.6 branches should probably be run through Travis as well.)

[Francois-Xavier Le Bail]

Post by Guy Harris

Post by Francois-Xavier Le Bail
https://gist.github.com/fxlb/10525029

Yes, those branches are branches that exist in the bpf.tcpdump.org repositories;
I guess pushing from there to GitHub causes the GitHub repository to have them
as well.
Most of them are not worthy of being run through Travis, at least as far as
I'm concerned; can it be configured to build only the trunk (master) for
now?  (I think the next releases will be 1.6/4.6, coming from the trunk; once
those are branched, the 1.6 and 4.6 branches should probably be run through
Travis as well.)

1) Are the branches marked "merged into master" (tcpdump: bare and origin ; libpcap : bare and origin) still useful? Otherwise, they could be deleted.

2) For all branches, first question is : will new commits be pushed into this branch ?
if no : no more Travis problems.

3) The other question is: is there a '.travis.yml' file in the branch ?
if no :
a) No correct builds.
b) We cannot use the 'except' keyword for blacklist branches, nor 'only' keyword for whitelist branches.

So remain branches which have to evolve and have no '.travis.yml':
a) We need to add a '.travis.yml' file in the branch.
b) If they must not be build by Travis, one of the two aformentioned keywords could be used.

Same for new branches.

Hope this help.

[Francois-Xavier Le Bail]

Post by Francois-Xavier Le Bail
1) Are the branches marked "merged into master" (tcpdump: bare and
origin ; libpcap : bare and origin) still useful? Otherwise, they could be
deleted.
2)  For all branches, first question is : will new commits be pushed into this
branch ?
if no : no more Travis problems.
3) The other question is: is there a '.travis.yml' file in the branch ?
a) No correct builds.
b) We cannot use the 'except' keyword for blacklist branches, nor
'only' keyword for whitelist branches.
a) We need to add a '.travis.yml' file in the branch.
b) If they must not be build by Travis, one of the two aformentioned keywords
could be used.
Same for new branches.

I forgot the case of branches which have to evolve, have the '.travis.yml' file and
must not be build by Travis : one of the two aformentioned keywords could be used.

[Guy Harris]

Post by Francois-Xavier Le Bail
1) Are the branches marked "merged into master" (tcpdump: bare and origin ; libpcap : bare and origin) still useful? Otherwise, they could be deleted.

Michael? What are those branches for?

Post by Francois-Xavier Le Bail
2) For all branches, first question is : will new commits be pushed into this branch ?
if no : no more Travis problems.

At this point, I don't expect any new commits into any branches other than master.

There will probably be libpcap_1_6 or libpcap_1.6, and tcpdump_4_6 or tcpdump_4.6, branches once the 1.6 and 4.6 releases are made, and there might be commits into those branches.

[Francois-Xavier Le Bail]

Post by Francois-Xavier Le Bail

  2)  For all branches, first question is : will new commits be pushed into

this branch ?

  if no : no more Travis problems.

At this point, I don't expect any new commits into any branches other than master.
There will probably be libpcap_1_6 or libpcap_1.6, and tcpdump_4_6 or
tcpdump_4.6, branches once the 1.6 and 4.6 releases are made, and there might be
commits into those branches.

I don't see any problem with these branches. They will get their .travis.yml from master and Travis will work accordingly.