OPML | The Coffeehouse

Rossj

Robert has taken to chiding people for moaning about the quality of the OPML spec, rather than building an app that works with someone else's implementation of OPML.

Personally I don't think it is the spec that is broken, rather the format itself, relying too much on word of mouth agreement on the meaning of certain XML attributes in the format. The type attribute in particular, for an RSS feed, should it be rss? text/xml? url? what about text? text? string?

So what do you think? Is word of mouth agreement on parts of an XML format a good idea if you want to build support for a format? Or do you think developers should be spending all their time working out how someone has decided to implement a particular XML dialect and 'guess' the attribute values?

Personally I'd like to see some XSD - but that's just me Smiley

Manip

Excuse me for trolling but isn't XML supposed to solve this exact problem?

I love text formats, I think they are great. Everything is a UNICODE string, new lines split records and parts of records might be split using special characters (e.g. '='). Best of all they are super-quick to process.

With any text based format you have to document 100% of the format. I mean if you don't tell someone that every new line is a \r\n then who is going to? And if you don't tell them a comment is // then how are they supposed to know?

Now, XML came along.. You have lots of "experts" shouting about how you no longer have to document because XML is self-documenting and such... They also say that because it is more structured less documentation is needed and thus it is more cross platform, or something like that. But as I said at the time none of that is even slightly true.

All XML really does is slows down the application, and makes the files harder for humans to read and edit. With XML, just as with text files, if you don't document it 100% then others are going to badly implement.

Side Note: I really don't care what Scoble thinks about OPML... He isn't qualified to have an opinion on the subject.

jamie

Manip wrote:

Side Note: I really don't care what Scoble thinks about OPML... He isn't qualified to have an opinion on the subject.

Why cant a user have an opinion?

Manip

jamie wrote:

Manip wrote:

Side Note: I really don't care what Scoble thinks about OPML... He isn't qualified to have an opinion on the subject.

Why cant a user have an opinion?

I didn't say a user can't have an opinion, I said I don't care what the user's opinion is because they aren't qualified to hold one of any value on this subject.

jamie

why though?

users have different perspective than devs

thats good

Black Ratchet
Just another Phone Phreak from Boston

Manip wrote:
Side Note: I really don't care what Scoble thinks about OPML... He isn't qualified to have an opinion on the subject.

blowdart
Peek-a-boo

jamie wrote:
why though?

users have different perspective than devs

thats good

And until developers realise that software will suck. Hard.

Manip

User's opinions about file formats is not important. The graphic interface, the functionality and such topics are absolutely the realm of users to complain about. But one developer complaining about another developer's documentation on a file format is completely none of the user's business.

billh
call -141

But it becomes the user's business when developer infighting leads to poor UI implementations, or conversion messes. Then, it goes right back to the developer.

Rossj

My argument is that users should only care about what happens 'under the hood' when it is doing something to stop them doing what they want. If the fact that developers can't agree on a format because there is no documented standard to follow then that impacts the user.

The user doesn't need to worry that developers are arguing about the file format, except that those who are arguing that we can ignore standards will be the first to complain when apps start not being able to interoperate, or only work in some places.

As far as developers are concerned it comes down to this

opml spec wrote:

type is a string, it says how the other attributes of the <outline> are interpreted.

Unfortunately there is no further mention of what the type should actually contain. Some developers will put a type of text for a string, others string, others words, others something else. If I have to make *assumptions* on what these values could be, then I am likely to miss out one of these potential values or miss one that is used in the future. This will make things break. This will not impress users.

I've already suggested (as have many others) that if the spec said this should be a mimetype then it would go a long way to ensuring that a user exporting OPML from one app will not have problems when importing to another - as is currently the case. In fact go read the spec, and tell me as a developer whether it is clear enough for you to write software that uses it and not cause problems for users further down the line.

Saying - just do what Dave does stinks of 'old Microsoft', you know before they started trying to be transparent and more friendly.

*Nothing* goes in my software unless there is a direct benefit to the user (and often it is only the user that can tell me that) so I am not being 'anti-customer'.

Charles
Welcome Change

Let's not forget that developers are also users. In fact, User is the base class for all types that rely on software to get things done, explore the virtual world, communicate in a digtal way, etc. Scoble has just as much of a right to complain about OPML specifications as Don Box or Chris Anderson. Of course, the context will be different since Scoble doesn't write code, but the issue is outcome, not strictly implementation.

Power to the user,
C

Rossj

Thing is, Robert isn't complaining about the spec, he is defending it - he is railing at all those developers that dare complain that they can't code to a badly put together spec without upsetting users, whilst Robert just wants us to all just copy whatever it is Dave W is doing.

It's all part of a developers don't care about users conspiracy Embarassed

billh
call -141

I also have an issue with this line:

Scoble wrote:
But, he does something most of the other, more superior, spec writers rarely do: he wrote an app.

Maybe nowadays there is a bigger disconnect between the spec writers and the app writers, especially with all the standards committees that abound for one thing or another...that wasn't always true in the past, though. Plus, it really does depend how many users this "spec" will impact...are we talking 5, 10, 100 or 1,000,000 users?

But I'll make a bigger comment over there in his mudpit where it belongs.

Charles
Welcome Change

Rossj wrote:

It's all part of a developers don't care about users conspiracy

Let's dispell that myth once and for all! Again, developers ARE users. We don't like to use apps that suck as much as non-developers don't.

C

JChung2006

I don't think it's that developers don't care about users so much as it is that developers can be out of touch with users.

Certainly, users can be out of touch with developers since they are usually unfamiliar with what sort of issues developers face in the production of software. However, developers can be just as out of touch with users since they can be unfamiliar with the usage of software from the users' perspective.

Charles
Welcome Change

Certainly, unusable software is useless. All good developers must always have users in the fore of their thinking when designing software. Implementation details are developer specific for sure, but so too should the usability of the software they are creating.

Historically, you are correct that developers tend to forget about the usage patterns of their software (it's not always easy to predict how people will actually use your software). The times, however, are a changin'.

C

blowdart
Peek-a-boo

Charles wrote:
Certainly, unusable software is useless. All good developers must always have users in the fore of their thinking when designing software. Implementation details are developer specific for sure, but so too should the usability of the software they are creating.

My wife despises training word because of the toolbars. She's firmly of the opinion that the default toolbar should have a maximum of 10 icons on it, the bacis that you see in the rich text box here, bold, italics, underline, paragraphs, bullets. That's all 90% of the office workers she trains need.

Don't ask her about the new UI either Perplexed

Minh
WOOH! WOOH!

Manip wrote:
I love text formats, I think they are great. Everything is a UNICODE string, new lines split records and parts of records might be split using special characters (e.g. '='). Best of all they are super-quick to process.

How would you handle hierarchical data?

Manip wrote:
Side Note: I really don't care what Scoble thinks about OPML... He isn't qualified to have an opinion on the subject.

That never stopped you before! Haha Big Smile

scobleizer
I'm the video guy

Rossj wrote:
Thing is, Robert isn't complaining about the spec, he is defending it

You are misreading me. I never defended the spec. I defended the implementation. The two are different things. Maybe I didn't make myself clear. I made it very clear what I wanted (hint: it's already implemented so all you need to do is copy what's already been done). I just wanted a feature added to that. To take my feature request and then start arguing about the quality of the spec is simply bad listening.

OPML is both an application and a format. I really don't care how you implement what I want. But the OPML APPLICATION is the first time I've been given what I want by a developer. So, for another developer to attack it is funny to me.

Developers sure aren't very good at listening to users.

And, note, neither are we. That's why it took years to update Sound Recorder in Windows.

That all said, if you have a better implementation because you have a better spec and a better format already done, let me know. I'd love to use it!

Rossj

Robert, developers are not attacking the app, they are attacking the assumption that they have to follow a way of working with a protocol that is badly defined. Why? Because of this bad definition that developers are worried that there will be issues with compatability at some later point - previous uses of the protocol have already shown that. It isn't an attack on Dave, the spec really is ambiguous enough to be dangerous if you were relying on it for any sort of future scalability. Sure both RSS and OPML (spec) work, but how many hours have been wasted by people trying to work around the ambiguity in the specs when they could have been working on something else? Dave's big invention is not cool because of the format, it is cool in spite of it. I still haven't seen Dave's 'extensions' to OPML that would be needed to support the app, any idea where I can get it - or do I have to trawl through megabytes of source code looking for it?

Rather than asking developers to 'copy that' how about putting in a bit of effort - after all you'll probably get what you want for free, so how about rather than just saying 'I want that' you tell us what the problem is and let us *have a conversation* to determine the best solution. To paraphrase Tom from the mudpit, tell us what you want, not how to do it. You might be surprised how many developers DO care about what the users want.

Developers need to feel that their app is going to be useful (although I've seen plenty of examples that suggest otherwise) and that a user is going to enjoy using it, or at least be able to use it to make a specific task easier. Maybe that's just the mac fanboy in me Smiley

And Robert, please, whatever happens, don't think you have to defend anything Dave says or does because he is a friend, he's a big man and can stand up for himself.

Rossj

I forgot to mention, and can't edit my posts in Firefox.

This whole - developers are useless at giving users what they want, I've seen it before, almost verbatim, from Winer. And it's bull.

Manip

If you gave users everything they want all the time you would never ship. Users clearly are getting SOME of what they want otherwise how would you explain success stories like Windows and Google?

dentaku

If you give users everything they want you would have a screen full of buttons and toolbars.
That's usually my response to old (mostly old) people who ask "why isn't there a button right there that does what I want".

It reminds me of the old Lightwave 3D plug-in called Eagle in a Barrel. You clicked one button and it created a 3D model of an eagle in a barrel. It was a completely useless feature but I guess it was there as a statement to show how people always want a simple one button solution to complex tasks instead of using the tools they already have to arrive at the same result even though it involved a little more work.

Maurits
AKA Matthew van Eerde

Rossj wrote:
The type attribute in particular, for an RSS feed, should it be rss? text/xml? url? what about text? text? string?

application/rss+xml seems to be the type of choice.

AndyC

Rossj wrote:

Is word of mouth agreement on parts of an XML format a good idea if you want to build support for a format?

Anyone who thinks it is should go back and look at the storm created by MSFT's implementation of Kerberos. It follows the spec, it just doesn't follow all the word of mouth agreements that never made it back into the spec.

If OPML is to succeed Dave needs to spend a couple of hours clarifying the format a little. Then Scoble can have his features and share files with all the people he wants to without any of that unpleasant hassle!