Skip site navigation (1)Skip section navigation (2)
Date:      Sat, 4 Jul 2015 12:40:25 +1000
From:      Kubilay Kocak <koobs@FreeBSD.org>
To:        "Sergey A. Osokin" <osa@FreeBSD.org>
Cc:        ports-committers@freebsd.org, svn-ports-all@freebsd.org, svn-ports-head@freebsd.org
Subject:   Re: svn commit: r391254 - in head/www: nginx nginx-devel
Message-ID:  <55974799.8050907@FreeBSD.org>
In-Reply-To: <20150703231507.GC24716@FreeBSD.org>
References:  <201507031644.t63GixME014247@repo.freebsd.org> <20150703172909.GB24716@FreeBSD.org> <5596CE3C.5000801@FreeBSD.org> <20150703231507.GC24716@FreeBSD.org>

next in thread | previous in thread | raw e-mail | index | archive | help
On 4/07/2015 9:15 AM, Sergey A. Osokin wrote:
> On Sat, Jul 04, 2015 at 04:02:36AM +1000, Kubilay Kocak wrote:
>> On 4/07/2015 3:29 AM, Sergey A. Osokin wrote:
>>> Dear Kubilay,
>>>
>>> I didn't approve this change, so, I have at least two questions here:
>>
>> I believe the tag was moved 11 days ago and the issue (PR) created 7
>> days ago. A number of users had reported the issue today as ongoing,
>> which is when I found the bugzilla issue. I had assumed you weren't
>> otherwise available and wanted to help.
> 
> Have you asked those users to add "+1" to PR 201129?

I did not, but I will next time. Would this have made my unapproved
commit more OK?

>>> 1. have you checked what actually has been changed?  Is there any chance to see 
>>>    the diff between old distro and new one?
>>
>> I did not, I considered it the same as I would have a normal version
>> bump of a module, except in this case the distinfo checksum mismatch was
>> caused by upstream moving a tag, not a maintainer forgetting to run makesum.
> 
> Well, I don't think that this is good idea to commit every change to the "super popular
> software packages", what I've heard in the PR.

The change in question was only to address a two week broken build. I
certainly agree, and do not subscribe to 'committing every change'
outside of normal processed. Re-rolled distfiles and re-tagged git
repositories are a regular occurrence, with a standard fix.

> The size/SHA256 mismatch in third-party headers_more module has been acquired probably
> because of the module's author mistake (but I think he thought he did his best): he's
> changed something in source code after the creation of the release tag.
> 
> Another version a bit paranoid, but anyway: somebody hacked a github account, committed
> a troyan, re-created the tag and Kubilay added that troyan into FreeBSD ports tree.
> Actually this is why I'm asking you to show the changes between versions.

Agree with you on the potential risks. I did check commit history, I
didnt review code.

> In my point of view, I'd highly recommend to ask the module's author about change,
> create new release with that change.

feld@ created the following issue, to which I've added a comment
requesting documentation of the changes:

https://github.com/openresty/headers-more-nginx-module/issues/35

>> I don't know how to see how/where a tag was moved between commits, so as
>> to derive a changeset.
>>
>> It would be nice to know if there is a way.
>>
>>> 2. the third-party headers_more undefined by default, so PORTREVISION bump
>>>    isn't necessary in this case.
>>
>> Understood. I had originally thought that since the distinfo was
>> packaged, and that the contents was changing, that it may have been
>> required.
> 




Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?55974799.8050907>