Opened 4 years ago

Closed 3 years ago

#813 closed Bug / Defect (fixed)

Travis and Buildbot builds are broken due to MacOS X + cmocka issue

Reported by: Samuli Seppänen Owned by: Samuli Seppänen
Priority: major Milestone:
Component: Building / Compiling Version: OpenVPN git master branch (Community Ed)
Severity: Not set (select this one, unless your'e a OpenVPN developer) Keywords:
Cc: chipitsine@…

Description


Change History (15)

comment:1 Changed 4 years ago by Samuli Seppänen

Cc: chipitsine@… added
Owner: set to Samuli Seppänen
Status: newaccepted

comment:2 Changed 4 years ago by Gert Döring

I've sent suggestions how to tackle this in configure.ac to the list - so "someone needs to look at them, see what is doable, and send a patch".

comment:3 Changed 4 years ago by chipitsine

there are at least two persons responsible in that breakage: the commiter and one who "ack"ed. I think it makes sense to expect fix from those persons

comment:4 in reply to:  3 Changed 4 years ago by Gert Döring

Replying to chipitsine:

there are at least two persons responsible in that breakage: the commiter and one who "ack"ed. I think it makes sense to expect fix from those persons

There seems to be a slight misunderstanding on your side about the our ability to make "those persons" provide a fix? This is volunteer work. Who finds something annoying comes up with a fix, or politely asks someone more capable to do the fix.

comment:5 Changed 4 years ago by chipitsine

if nobody provides a fix, it makes sense to revert that commit.

comment:6 Changed 4 years ago by chipitsine

I think, reverting should be done right after travis-ci got broken.

comment:7 Changed 4 years ago by Gert Döring

No. The commit brings in test code that provides important unit test code for the argv rewrite - this is good. That it breaks MacOS X compilation under certain circumstances (namely: git submodule is checked-out and cmake is installed) is an annoying side effect, but not important enough to revert the commit.

This is the way we work: important changes get made, and if they break some niche thing, someone fixes that - sometimes the original committer, sometimes someone else who happens to have an interest in fixing it.

comment:8 Changed 4 years ago by chipitsine

travis-ci provides automatic testing for commits. currently it is broken for months.

if you do not see the importance of automated testing, why have it at all ?

as for fixing, someone made that commit, it is responsibility. someone "ACK"ed it, it is also responsibility.

I see no activity from those persons to resolve it. what does it mean ?

comment:9 in reply to:  8 ; Changed 4 years ago by Gert Döring

Replying to chipitsine:

travis-ci provides automatic testing for commits. currently it is broken for months.

... broken *for MacOS*.

if you do not see the importance of automated testing, why have it at all ?

Travis-CI seems to be something important to you, and this is why we have it.

We test with buildbot and t_client setups, and indeed, the MacOS X buildslave is a bit of a nuisance these days (unless Samuli already fixed it). But buildbot tells me "all others are green" so I can live with the failing ones.

as for fixing, someone made that commit, it is responsibility. someone "ACK"ed it, it is also responsibility.

I see no activity from those persons to resolve it. what does it mean ?

They are busy?

comment:10 in reply to:  9 Changed 4 years ago by chipitsine

Replying to cron2:

Replying to chipitsine:

travis-ci provides automatic testing for commits. currently it is broken for months.

... broken *for MacOS*.

if you do not see the importance of automated testing, why have it at all ?

Travis-CI seems to be something important to you, and this is why we have it.

We test with buildbot and t_client setups, and indeed, the MacOS X buildslave is a bit of a nuisance these days (unless Samuli already fixed it). But buildbot tells me "all others are green" so I can live with the failing ones.

as for fixing, someone made that commit, it is responsibility. someone "ACK"ed it, it is also responsibility.

I see no activity from those persons to resolve it. what does it mean ?

They are busy?

I think we can ask them are they busy or ... ?

comment:11 Changed 3 years ago by Gert Döring

Patch sent to list. Wasn't my commit or ACK, but I think just coding the fix (two-liner) is spending my time much better than pointing fingers at "who should do this".

https://www.mail-archive.com/openvpn-devel@lists.sourceforge.net/msg13843.html

comment:12 Changed 3 years ago by Samuli Seppänen

@cron2: oh, but it's all about the principle :). Anyways, let's hope that the patch fixes the problem!

comment:13 in reply to:  12 Changed 3 years ago by Gert Döring

Replying to samuli:

@cron2: oh, but it's all about the principle :). Anyways, let's hope that the patch fixes the problem!

It seems to, but dazo thinks it's not the right way to do it. So I'm waiting for someone else to bring forward a better patch.

comment:14 Changed 3 years ago by chipitsine

when dazo is going to send new patch ?

comment:15 Changed 3 years ago by Gert Döring

Resolution: fixed
Status: acceptedclosed

commit f91ab283a407e25c4b32aecb390911b212ce2694
Author: Selva Nair <selva.nair@…>
Date: Wed Jan 18 15:42:52 2017 -0500

Add a check for -Wl, --wrap support in linker

Note: See TracTickets for help on using tickets.