[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: Philip Martin <philip.martin_at_wandisco.com>
Date: Wed, 09 Mar 2011 09:40:00 +0000

Greg Stein <gstein_at_gmail.com> writes:

> On Tue, Mar 8, 2011 at 11:08, Philip Martin <philip.martin_at_wandisco.com> wrote:
>> VTXN operation:
>>
>>  - client sends POST
>>  - proxy adds SVN-VTxn-Name:UUID
>>  - server creates transaction called TXN-NAME
>>  - server replies SVN-VTxn-Name:UUID
>>  - proxy passes
>
> or:
>
> - proxy adds: SVN-VTxn-Name:UUID
>
>>  - client sends !svn/vtxn/UUID
>>  - proxy passes
>>  - server extracts UUID and maps to TXN-NAME
>
> So my question is whether to have the server do it, or have the proxy do it.
>
> Basically, the server is responding with somebody the requestor
> already knows. So I wonder which approach is "best". It seems to be
> kinda six-of-one/half-dozen-of-another. I suspect the server just
> needs an if/else, so that might not be nearly the burden relative
> modifying the proxy to add that header.

The server needs an "if" in order to avoid sending TXN-NAME when it
receives VTXN-NAME (and not sending TXN-NAME is the main reason for this
change). If the server doesn't send VTXN-NAME then the response is not
complete and the server is relying on the the proxy to make it complete.
I think it is better if the server generates a complete response.

-- 
Philip
Received on 2011-03-09 10:40:41 CET

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

This site is subject to the Apache Privacy Policy and the Apache Public Forum Archive Policy.