In IE8 Beta 1 we also increased our per-server connection limit from 2 to 6. What this means is that in IE7 and below pages could only download 2 elements from a given server at any one time. Increasing that limit to 6 allows sites to download 3 times as much content in parallel, which should translate into faster page download times when bandwidth is available.

And it will. Today web pages contain many more references to stylesheets, images and external javascript files than they used to. Increasing the number of parallel connections will make pages complete more quickly on the user side. However, the HTTP protocol specification says this about concurrent connections:

Clients that use persistent connections SHOULD limit the number of simultaneous connections that they maintain to a given server. A single-user client SHOULD NOT maintain more than 2 connections with any server or proxy.

Granted, the HTTP specification was written a long time ago; internet connections and servers have become much more powerful. So, will it matter?

My main problem is that the decision has moved from server administrators to browser developers. If you are a server administrator it is easy to split up content so that it appears to be coming from multiple hosts to get the same result. You know more about the content that needs to be delivered and how it should be partitioned.

If IE8 uptake is quick I guess some popular sites will see some negative effects of the 300% increase in the number of connections.

I have a 24 Mbit broadband connection but it will still take some time to download 180 Mb. If I close the lid on my laptop or loose the connection for a short while the download will restart from scratch. It is 2008 now and Content-Range operations in HTTP 1.1 has been around for quite some time.

At the heart of the issue is the lack of reference implementations. ISO is way behind W3C in this area. Could someone please tell ISO that open source reference implementations are an absolute necessity when working with standards for information exchange?

From the W3C technical report development process section 7.4.4:

Preferably, the Working Group should be able to demonstrate two interoperable implementations of each feature.

It is simple really. The benefit of a standard is created when it is used. Open source reference implementations shortens the time to market for everyone implementing the standard in their products and also disambiguate interpretation of the standard specification.

Tim, please tell me you know someone at ISO that can fix the process.

Seriously? Can’t Microsoft get their own implementations to cooperate better? And this has just been approved as an ISO standard?

Said the monk:

If you give me non-standard markup, I will render it according to standards.

If you give me standard markup, I will not render it according to standards.

What do you do?

The student sat for a long time and said nothing. Then, without looking up, he raised one finger and said, “There is only one web.” Many years later, the monk was enlightened, but by then it was too late.

I thought the whole idea is that a standard is a contract that tool makers and content producers should be able to rely on. And now you are saying that the standard isn’t enough but that I specifically must inform a particular browser that I want standards standards mode?

Also see Ian Hixie’s entry Mistakes, Sadness, Regret.

The member countries have had six months to consider if the Office Open XML (OOXML) format should become an ISO standard. In Sweden, SIS arranged a working group that have looked through the material. As you may know the OOXML format has been heavily criticized (by many e.g. Google (PDF)) for allowing embedding of closed Microsoft-specific objects in the document standard and thus making it difficult for non-Microsoft software to read OOXML documents.

Unfortunately, SIS is an organisation where anyone can become a member. Member organizations can send participants to a working group for a fee. The current rate is 17,000 SEK (~$2,500). The day before the vote that decided if SIS would say yes to OOXML in the ISO there were a couple of new members in the SIS/TK321/AG17 working group:

And so, Sweden will be voting yes to make OOXML an ISO standard.

For more information see:

• Patrik Fälström’s post “Microsoft managed to buy the vote of Sweden in ISO?“.
• Slashdot.
• Dagens Nyheter (in swedish).
• noooxml.org.
• Matusow: Open XML – The Vote in Sweden

.