[svn.haxx.se] · SVN Dev · SVN Users · SVN Org · TSVN Dev · TSVN Users · Subclipse Dev · Subclipse Users · this month's index

Re: [PATCH v2] HTTPv2 allow client to control transaction name in protocol

From: Greg Stein <gstein_at_gmail.com>
Date: Tue, 8 Mar 2011 10:47:40 -0500

On Sat, Mar 5, 2011 at 13:16, Philip Martin <philip.martin_at_wandisco.com> wrote:
> Greg Stein <gstein_at_gmail.com> writes:
>
>>> If the client sends, or a proxy injects, an SVN-VTxn-Name header with
>>> the POST request it defines the transaction name to be returned to the
>>> client in the POST response.  If the client recieves the new
>>> SVN-VTxn-Name header it uses that name in the new URIs in the requests
>>> that make up the commit.
>>
>> I don't understand why the *client* needs to read that header. The
>> base URI that the server returns already has the proper txn name,
>> right?
>>
>> Sending *to* the server creates the "client-provided" feature that you
>> want to retain. I just don't understand the other direction.
>
> By default the client doesn't send the header, it receives SVN-Txn-Name
> containing the server generated name and uses that in the txn/txr URIs.
> A client could send SVN-VTxn-Name but most clients will not.  If a proxy
> inserts an SVN-VTxn-Name header into the request, the server replies
> with an SVN-VTxn-Name header containing the proxy generated name.  The
> client then receives SVN-VTxn-Name containing the proxy generated name
> and uses that with the vtxn/vtxr URIs.

Ah. Gotcha. Thanks.

And to be clear: the server *could* just remain silent, and the proxy
would insert the SVN-VTxn-Name header in the response back to the
client, right? Would that be an improvement/simplification?

Cheers,
-g
Received on 2011-03-08 16:48:10 CET

This is an archived mail posted to the Subversion Dev mailing list.