[Date Prev][Date Next] [Thread Prev][Thread Next] [Date Index] [Thread Index]

Re: Intent to hijack Bacula (Heads up, Get The Facts!) (long)

John Goerzen wrote:
> Hello,
> I intend to take over the Bacula package.  I would first like to say
> thanks to Jose Luis Tallon for initially packaging it for Debian and
> maintaining it for these years.
You have a funny sense of time, don't you?
This is true; Years. Since October 2003.
> A brief history of why I intend to do this:
>  * Bacula has had RC bugs open for more than a year.  It was removed
>    from testing several months ago because of this.
>  * Bacula's current maintainer is not a Debian Developer and has been
>    in NM since 2003.
So what? If only the "elite" can contribute to Debian say so. Then, all
non-DD maintainers will quit;
We might as well flee away to Ubuntu like many former Debian users have
done and are doing.
At least, they manage to get a distro out of the door every six months.

Note, however, that I love Debian as a distro. I just hate the rudeness
of some people here (and regret that most of the gentlemen DDs that I
have encountered in these years very seldom raise their voices here -- I
assume they are devoting that time to Debian in some other way)
>  * Bacula as it currently exists in sid is unbuildable and
>    uninstallable.  Bacula will not be present in etch unless significant
>    problems are fixed.
There is still half a year left until Etch is released.
Still much more than two months left for the base freeze. A transition
takes 10 days at most.
>  * The last upload for Bacula was almost a year ago.
There were no upstream releases for over six months, either.
>  * The maintainer has repeatedly, over the last year, said he's working
>    on this but hasn't made much real progress, and has made no upload to
>    Debian.
And I have. Prove otherwise if you can. I have my testing standards, and
never upload anything without testing.
You blame me for "not testing packages and
>  * Several additional critical-level or grave-level unreported bugs
>    exist in the bacula Debian source tree (such as stopping database servers
>    without permission and deleting files un-owned by a particular
>    package)
Then report them. That's what the BTS is for.
>  * There are various policy compliance issues with the current packages.
So? report them. That's the procedure you are supposed to follow.
>  * The current maintainer does respond to pings, but has a long record
>    of problems getting bugs (even RC bugs) fixed in a timely fashion.
I assume you meant "not fixed"...
> I have already prepared an NMU that fixes 22 bugs, including all four RC
> bugs.  I have tagged those bugs as pending.  This release is currently
> sitting in NEW.  I also prepared subsequent NMUs that fix critical, but
> unreported, bugs in the Debian Bacula packages.
The fact that you uploaded six versions of Bacula to NEW within one day
gives an idea of the level of testing you give them.
For those unaware of Bacula's packaging: there are several components,
in three flavors (SQLite, MySQL, PostgreSQL).

This means that John Goerzen uploads packages without testing them.
Either that or he enjoys 48h-days and has ultra-fast machines (I have an
AMD3GHz/1GB RAM/SCSI discs for development).

JG>It is not your place to upload known-broken software to sid.  It is also
not your place to fail to test them before uploading.

Please tell me how in hell can you justify accusing me of not testing my
packages, when you have obviously not done so.
You seem to have some fixation with uploading, don't you?? Six versions
in 24h ?
Instead of uploading many half-baked packages, you could try getting one
right before uploading.
It is hardly justifiable to build and upload that many packages if you
already foresee many more fixes needing to go in.

> Fixing the rest of the problems with Bacula requires a level of work
> that is not really appropriate for an NMU.  I have discussed the
> situation with Jose's AM, Stephen Frost, who encouraged me to hijack the
> package.
Stephen Frost: Thank you for showing how an AM *should not* proceed.
After neglecting me for several months, and failing to process my NM
answers in due time, you have now joined John Goerzen in trying to
discourage me from working for Debian.
I had to publicly ping you several times on -devel (as well as doing so
privately) before I could even get acknowledgement that you had received
my answers.

If you are that busy (as you have admitted on -devel), then step down as
AM and leave space for others who will be able to do better. You didn't,
yet you now blame me for not keeping up to my duties as a maintainer and
not wanting to work on this.
'nuff said.

(This doesn't feel impartial with me anymore -- that would disqualify
you as able to evaluate my work)
> I have also discussed the situation on #debian-devel, and consensus there seemed to be that a hijack was warranted.
> In short, I will be maintaining working Bacula packages 
My users don't seem to agree with you (w.r.t. to "working" Bacula
packages)... I have *many* successful reports from my users, by the way.
Wanna check?
> anyway, and I might as well upload them to Debian.
> I also e-mailed Jose about the situation, offering to adopt Bacula, on
> April 27, but have not yet received a reply to that message.
Well.. I was away, on vacation and without connectivity, for some days.
Am I required to always check my mail daily??

I answered as soon as I was back. Note that, not being a DD, I can't set
the "vacation flag" on db.debian.org.
Please point where it is said that I must have a "vacation
autoresponder" otherwise.

JLT>Moreover, i have had several personal issues during the last months,
which made me unable to help.

John Goerzen>Unfortunate, but the proper thing to do in this situation is to announce
the situation and solicit NMUs from people.  As far as I have seen, you
did not do this.

That's what the "request for help" tag is for. I did set it up quite some time ago, yet nobody cared until now..

John Goerzen>"he has written very poor code -- some of which has been broken for several years "

If the oldest bug is less than a year old, how could that be? Troll.

John Goerzen>"-- and has not made much effort to fix it."

You are not entitled to judge that, not having proof. A lack of uploads can mean a lot of things.

John Goerzen> "If Jose were primary maintainer, and I was a co-maintainer (or one of
several), he would still get the final say over whether or not he would
accept my patches, and would still get the final say over making changes
to the package.  I do not have confidence that he would bear that responsibility well."

You don't seem to be like working in a team. Do you? I have proved otherwise.

Quoting John Goerzen: "We intend to use it in a production environment where it will be backing up approximately 5TB daily and drive a 48-tape library.  I will be using work time to maintain it."

You have yet to prove that your *sudden* interest in becoming Bacula's primary maintainer is only for technical reasons. The fact that the company you work for is going to migrate to Bacula indicates that you might as well earn some points with you boss or what-have-you.

Prove me wrong if you can.

In facts: 
- John's first documented contact with Bacula was on March 21st 2006[1]. A series of postings to bacula-users (to which I am subscribed for three years already) asking very basic questions -- he was evaluating whether he could use Bacula to replace Amanda in his environment.

- In his own blog[2] he admits being a complete "Bacula newbie".

- He is not even subscribed to the bacula-devel list

- Has no contact with any of the users AFAIK.

... yet he pretends to know the software and its users better than me and so be able to redo the packaging in a better way in no time.

He never answered this question either:
JG> BTW, if you  feel so proud of being an "official" Debian Developer, why
don't you use your @debian.org account, not even for your packages? I surely would.

... yet he makes a big point of the fact that I'm no DD.

JLT>I was "reprehended" by Turbo Fredriksson due to the amount of
CPU wasted. He cared to contribute some patches which, after being
integrated and enhanced --as much as i could-- by me, form the current
build system.

My original build approach (in 1.32f6) used the method you suggest. That means that I can do it "properly" (using your own terms).
However, Turbo and PMHahn (and some others, too), advised me to change everything.
Turbo himself warned me of a better way to do it on March 2nd/3rd 2004, and pointed me to #196802

I completely rewrote the build system based on their input and got it working. My sponsor, rover, also appreciated it (its speed, mostly).

This means that there are DDs which approve of this approach as well as there are some who disprove of it. It can't be that bad, then.

After deliberating for a while and consulting with some DDs, I decided to statically link the bscan binaries (among others) in order to reduce bacula-sd's need for dependencies and as the best solutio at the time. I kept that decision during 1.36's lifetime, and until PgSQL's transition made it impossible to keep (the FTBFS originated there)

I did agree with Stephen Frost on the need to migrate to a 'split' architecture, with bacula-sd-$flavor packages, and implemented that in my bacula-1.38.5 packaging (never released)

John Goerzen worked from a somewhat more evolved package which i had posted to Sourceforge to create his initial 1.38.8-0.1 package. I had already made the split then.
This doesn't mean that he has done a great job so far. I appreciate good work as much as anybody else.

This was a long mail, and took some time to compose. If I forgot any facts... well, I'll try to add those later.
I'd better devote some time to packaging instead -- I'm really tired of all this already.


[1] https://sourceforge.net/mailarchive/forum.php?forum_id=8833&max_rows=25&style=flat&viewmonth=200603&viewday=21
[2] http://changelog.complete.org/posts/485-Bacula.html

Reply to: