couchdb-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Jan Lehnardt <...@apache.org>
Subject Re: Transactional _bulk_docs
Date Fri, 06 Feb 2009 08:12:45 GMT
Hi Paul,

thanks for nailing the history of this.


Can we now please just wait for Damien's patch to arrive
to start discussion. This is a bad situation but I'd like to
go forward and I don't want to spend any more time on
the forensics of how things came to be. I hope it is now
pretty clear, that our modus operandi for some of the
fundamental work that Damien is doing is a little
controversial, but the PMC agrees it is good for the
Software and so far everybody else* agreed. We'll work
on communicating ongoing things with the community.

The technical issues will be talked about when we know
what Damien's patch does in detail. Thanks Antony for
explaining again what your use-case is. This information
will be a valuable addition for the discussion to come.

We're all in this to make really awesome software. Can
we get back to that?


Cheers
Jan
--
* By which I mean literally everybody, look at the
buzz CouchDB is creating in the industry.

On 6 Feb 2009, at 06:13, Paul Davis wrote:

> On Thu, Feb 5, 2009 at 10:02 PM, Antony Blakey <antony.blakey@gmail.com 
> > wrote:
>>
>> On 06/02/2009, at 6:20 AM, Chris Anderson wrote:
>>
>>> Antony, maybe it would help for you to explain just exactly what you
>>> wouldn't be able to do, without the bulk docs API. It will help to
>>> inform people about the technical issue.
>>
>>
>> My original email included this:
>>
>> -------------------------------------------------------
>>
>> For example, I have documents that can be cloned. The cloned document
>> contains a reference to the originating document. Then I delete the  
>> original
>> document, the clone history needs to be updated to remove the  
>> reference to
>> the original document and replace it with an original-deleted  
>> history item.
>> There is a business case that requires this consistency.
>>
>> With a transactional API this is easy. Without it, I can't see a  
>> way to
>> maintain consistency in the face of concurrent application access  
>> and/or
>> failure.
>>
>> -------------------------------------------------------
>>
>> However, I don't think this is really about a specific example.
>>
>> The problem is that if you get one side of the relationship written  
>> and
>> visible, but the other side not, then other concurrent accessors  
>> will see a
>> partially successful update.
>>
>> One response is "but you'll see this problem during replication",  
>> but I
>> think this is making a big assumption about how replication is
>> managed/interleaved with local application behaviour.
>>
>> Replication, and dealing with conflicts, is in no way automatic. As  
>> others
>> have stated, there is no domain-independent way of resolving  
>> conflicts.
>> Surely if it were possible to build a transactional API on top of a
>> conflict-based system, then this statement would not be true?
>>
>> I am deploying CouchDB like a Notes CLIENT. Not as a high-performance
>> database server. Replication is an explicit operation, that halts  
>> normal
>> activity. For my first delivery, replicas are read-only, so  
>> replication
>> conflict isn't possible, but when I move to a distributed writers  
>> scenario,
>> resolving replication conflicts will involve a specialized UI, that  
>> allows
>> all conflicts to be resolved before normal operation resumes. Thus  
>> the
>> editing application always sees a conflict-free database.
>>
>> The use-case of someone doing a local operation e.g. submitting a  
>> web form,
>> is very different than resolving replication conflicts. Conflict  
>> during a
>> local operation is a matter of application concurrency, whereas  
>> conflict
>> during replication is driven by the overall system model. It has  
>> different
>> temporal, administrative and UI boundaries.
>>
>> In short, I think it is a mistake to try and hide the different
>> characteristics of local (even clustered) operations, and  
>> replication. You
>> may disagree, but if the system distinguishes between these two
>> fundamentally different things (distinguished by their partition- 
>> tolerance),
>> you can code as though every operation leads to conflict if you  
>> wish, but I
>> can't take advantage of the difference.
>>
>>> I know that the long-standing vision of Couch doesn't include  
>>> special
>>> API exceptions for when you are running on a single node. And I'm a
>>> little afraid that the transactional doc commits Antony wants us to
>>> keep, are only a mirage, which would lead to trouble anyway, when
>>> distributed systems are involved.
>>
>> I don't understand why this needs to be the case. You can do  
>> transactions in
>> distributed systems. Do you have a model that isn't amenable to a  
>> Scalaris
>> treatment? Especially given that we're only talking about  
>> transactions over
>> a set of processes that are providing an illusion of a single  
>> system. Such a
>> cluster already requires some degree of partion-tolerance, right?  
>> And if
>> not, then what distinguishes a cluster from a partition-tolerant  
>> p2p mesh?
>>
>> Antony Blakey
>> -------------
>> CTO, Linkuistics Pty Ltd
>> Ph: 0438 840 787
>>
>> The fact that an opinion has been widely held is no evidence  
>> whatever that
>> it is not utterly absurd.
>> -- Bertrand Russell
>>
>>
>>
>
> I'm upset that CouchDB doesn't make me coffee in the morning.
>
> But the thing is, CouchDB is totally willing to make you coffee *and*
> bacon. It loves you *that* much.
>
> Enough with the silly. I've watched this drama avalanche for awhile
> and I finally think it's time for me to put out a few words on what
> I've seen.
>
> A brief history:
>
> 1. The mythical IRC conversation on 'removing' the feature: (roughly  
> quoted)
>
> Damien: I don't think we can support transactional commits in the face
> of multiple nodes. We can do ACID writes to disk so that updates
> aren't lost, but checking with an unbounded number of servers that a
> commit doesn't conflict isn't feasible.
>
> Everyone else: That's pretty reasonable.
>
> 2. A patch was applied to trunk that made commits to CouchDB
> optionally ACID compliant (which gives users the traditional
> speed/safety choice) as well as removing the atomic 'all or none'
> semantics.
>
> 3. Huge ML threads.
>
> History complete.
>
> Current status (through my eyes):
>
> Near as I can tell Damien has been nose to the grindstone for quite
> some time on this very specific part of the api. Would I like more
> status updates and ideas on where he's heading? Of course. Do I trust
> him? Yes. Is the community as a whole going to blindly accept some
> asinine patch that has no value that removes a crap load of
> functionality? No.
>
> Controversy!
>
> I tend to think that the 'discussion' that everyone keeps referring to
> hasn't even occurred yet. I look at the patch that was applied that
> caused this as an unfortunate early release.
>
> What?!
>
> Admissions first: I have no money riding on this issue. Whether or not
> CouchDB has transactional _bulk_docs worries me not at all. Though, I
> can't say that I have that much sympathy for a business model that
> relies on an open source project's trunk to remain compatible with
> required assumptions.
>
> Break:
>
> People seem to think that this conversation is over and done with. It
> isn't. This is a part of the API that's under work and will change.
>
> Reductio ad absurdum:
>
> Do we require a mailing list thread for every character changed in  
> the source?
>
> HTH,
> Paul Davis
>


Mime
View raw message