Improved viewer and script communications (Re: [sldev]
Puppettering Branch)
Teravus Ovares
teravus at gmail.com
Tue Jun 10 04:39:05 PDT 2008
*Tateru Nino*
Interesting, because I have four fully patched WinXP systems that don't have
the registry values at all. There's a discrepancy there, do you regularly
apply registry hacks?
Best Regards
Teravus
On 6/10/08, Tateru Nino <tateru.nino at gmail.com> wrote:
>
> I've just checked 3 WinXP systems (two with SP2 and one with SP3) and the
> registry settings for wininet for HTTP/1.0 and HTTP/1.1 connections are set
> to 0xa (10) on all of them.
> That's HKEY_CURRENT_USER\Software\Microsoft\Windows\CurrentVersion\Internet
> Settings MaxConnectionsPerServer and MaxConnectionsPer1_0Server.
>
> Teravus Ovares wrote:
>
>> Can anyone confirm that the client does not use a library that respects
>> this 2 connection limitation? So far in testing, it appears that it does.
>> When two threads get stuck, it fails to do anything else via http.
>> We've tried to use HTTP CAPS for inventory, and consistently, when the
>> inventory service runs slow, the client stops making *any* further http
>> requests.
>> Best Regards
>> Teravus
>>
>> On 6/9/08, *Tateru Nino* <tateru.nino at gmail.com <mailto:
>> tateru.nino at gmail.com>> wrote:
>>
>> Actually it is not a mandate. A mandate would be a MUST NOT. This
>> is a SHOULD NOT, specifically:
>> "A single-user client SHOULD NOT maintain more than 2 connections
>> with any server or proxy. A proxy SHOULD use up to 2*N connections
>> to another server or proxy, where N is the number of
>> simultaneously active users."
>>
>> I've got personal knowledge that the author did not intend the
>> above to apply to situations like this. For the substrate to MSIE,
>> however, it is entirely appropriate. Also, the above only applies
>> to persistent connections, not non-persistent connections
>> (applying the same guideline to non-persistent connections would
>> cause problems that this guideline is intended to avoid).
>>
>> Just because you're doing HTTP, doesn't make you a part of the
>> Web, and connection considerations in Web architecture over HTTP
>> are different to other architectures over HTTP.
>>
>>
>>
>>
>>
>> Teravus Ovares wrote:
>>
>> I also note, that according to Microsoft's kb article:
>> "The HTTP 1.1 specification (RFC2616) mandates the
>> two-connection limit. The four-connection limit for HTTP 1.0
>> is a self-imposed restriction that coincides with the standard
>> that is used by a number of popular Web browsers."
>> You can read the article here:
>> http://support.microsoft.com/kb/183110
>> You can read the RFC here:
>> http://www.faqs.org/rfcs/rfc2616.html
>> Best Regards
>> Teravus
>>
>>
>>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://lists.secondlife.com/pipermail/sldev/attachments/20080610/938f5e1e/attachment.htm
More information about the SLDev
mailing list