Good news...

[Phil Haack]

I was able to export my blog using BlogML. The reason it failed was due to a bug. Unfortunately BlogML has some serious shortcomings in the standard. Fortunately, it supports base64 encoding content, but I’m doubtful that other implementations support it. Of course, we do. J

Phil

[Phil Haack]

I forgot to mention the point of this. I think I’ll save the “long running blogml export task” work for the next version of Subtext. I really want to try and get this one out the door.

 

BTW, if there are small UI tweaks you think we should make, speak up nor or forever hold your peace (or at least hold it till the next version).

 

Phil

--

You received this message because you are subscribed to the Google Groups "Subtext" group.
To post to this group, send email to sub...@googlegroups.com.
To unsubscribe from this group, send email to subtext+u...@googlegroups.com.
For more options, visit this group at http://groups.google.com/group/subtext?hl=en.

[Simone Chiaretta]

We can always ask Keyvan to change the standard :)

> --
>
> You received this message because you are subscribed to the Google Groups "Subtext" group.
> To post to this group, send email to sub...@googlegroups.com.
> To unsubscribe from this group, send email to subtext+u...@googlegroups.com.
> For more options, visit this group at http://groups.google.com/group/subtext?hl=en.
>

--
Simone Chiaretta
Microsoft MVP ASP.NET - ASPInsider
Blog: http://codeclimber.net.nz
RSS: http://feeds2.feedburner.com/codeclimber
twitter: @simonech

Any sufficiently advanced technology is indistinguishable from magic
"Life is short, play hard"

[Phil Haack]

Well I have commit access too, but the problem is that the code and standard are so poorly written, how would any two different blog engines ever export/import to each other?

Let me give you just one example. Blog posts and comments both have a Content property of type BlogMLContent. The content can have one of several content types, "text", "html", "base64". When you write out blog posts and comments to the BlogML format, you'll get something like this, depending on the content type you choose:

<post><content type="text">...</content>
Or
<post><content type="base64">...</content>

So far so good, right? Here's where it gets screwy. When importing BlogML, you use the BlogML deserializer. Guess what, it can't deserialize that type. Why not? Because the type it deserialized to, BlogMLContent, does not have a content type property! In fact, it only has a Boolean base64 property, which is always false.

In fact, based on the types, it looks like what you're supposed to do is this when encoding base64 content:

<post><content type="base64" base64="true">...</content>

Which sees redundant. But that's not the only problem. The BlogMLWriterBase doesn't support writing the base64="true" attribute for comments and blog posts. You pretty much have to write your own writer and do it.

So here's the rub, there's two things I could do. Fix BlogML so it doesn't need the base64="true" attribute and instead it actually honors the type="" attribute correctly (there's a problem also with html encoded content). The problem with this is what if others are writing base64="true" instead of type="base64" (since that never worked).

Or I can make it write base64="true" properly, but what other blog engine out there will be able to import the content?

I'm tempted to simply drop the BlogML support altogether.

Phil

[Simone Chiaretta]

I think I have commit access as well.
Maybe we can get together with Keyvan and the other guy I don't
remember the name of to redesign the standard or at least make it less
prone to misinterpretations.
Or if the problem is the .net implementation of the parser, then we
can just fix it :)

simo

[Phil Haack]

Well I have a patch I prepared. I'll send it to Keyvan.

[Simone Chiaretta]

Good stuff :)