On Nov 10, 2006, at 12:25 AM, Mike Schinkel wrote:
> Looking from "the outside", respectfully speaking I think that is
> one of the
> problems with that which is called "REST." If you can't really
> describe it
> in anything other than abstract terms, and you can't really
> implement it
> without making it not REST, than does it really exist other than to
> stir up
> endless debate? Sorry if I'm being pedantic, but jeesh.
No, that isn't being pedantic -- it is being superficial.
What do you think abstractions are for? When your parents told
you that the leaves of a given tree are "green", did you have a
hissy fit over colors? Did you grab a leaf and demand that everyone
consider it to be the one true "green"? Did you declare it heresy
when the grass was also called green? Did you complain when the
crayon package came with various shades of "green"? Were you the
kid who stuck all the green crayons in his pocket, so that only
he could be the one true creator of the green fuzz-balls that made
brown rectangles look like trees? No?
Then why is it so hard to accept that an ARCHITECTURAL STYLE
cannot itself be implemented, just as there is no implementation
of green (only examples)? And that not being implementable in its
style form does not mean that an architecture cannot be constructed
in such a way as to fit our agreed style? And that being abstract
doesn't mean it is worthless to be able to talk about the style,
examine its characteristics, debate which ones are "best" for a
given application, and generally enjoy learning something that you
might not have known last week?
We can talk about green, various shades of green, and all the
other colors that, when combined, do not come out green. Children
have no problem understanding such things, though they might not
yet understand the properties of light and vision that make it all
work. Software architectural styles are much more complex, which
means you need to spend a lot more time thinking about the subject
(and drawing experiments) before it finally becomes clear. And
at that point you won't be complaining about abstractions: you'll
start looking for other colors.
... I don't care if it's called REST or "Walla-Walla-Hoo-Hoo." I'm looking pragmatically for something that adds a layer of value on REST; Frankly I don't...
... No, that isn't being pedantic -- it is being superficial. What do you think abstractions are for? When your parents told you that the leaves of a given...
Roy T. Fielding said RTF>> No, that isn't being pedantic -- it is being superficial. To-may-to, To-mah-to. RTF>> Then why is it so hard to accept that an...
... This is just not so. I wrote an article for onlamp about REST and AJAX. I *know* that other people have written similarly good articles (Prescod's stuff is...
Nic>> This is just not so. I wrote an article for onlamp about REST and AJAX. Wow. How do I say this w/o offending... Because of the organization you chose, I...
... [snip] ... Ordinarily, I don't participate so much in the list discussions - and I think it is worth my discussing why, in light of your statement above. I...
Alan Dead>> Hope that this illuminates :-) Yes, very helpful. I'll be interested in seeing what you offer. -Mike Schinkel http://www.mikeschinkel.com/blogs/ ...
... No, you just don't understand the terminology yet. You asked a bunch of gardeners for help implementing green, or at least an easy-to-read guide on doing...
Roy: I'm not here to attack you; I've got a tons of respect for you and what REST offers. Please read on... ... back in 1993. Yes, 1993. Since then he has...
... I agree with RTF that an "implementation guide" is a "different story" from the "architectural style" story that Roy told so well in his dissertation. So...
... Kind of, but not quite. They seem to be good examples but both seem to be a bit specific (i.e. not general purpose.) And they don't really seem to address...
... to be a ... Both claim to be quite general purpose: "The Google data APIs ('GData' for short) provide a simple standard protocol for reading and writing ...
... http://www.furl.net/item.jsp?id=12879797 . If GData ... I applaud that, but I'll be curious to see if they actually can use GData for everything w/o have...
... All that needs to be said is that REST is a design style by which implementations can be evaluated and system behaviours can be predicted. I can't evaluate...
... Nice and straightforward, but it does bring forward a clear question Bill. What tests would you apply to a system to assess it? Anyone? regards -- Dave...
... But my advice still stands, choosing singular vs. plural comes down to a matter of preference and not one of pragmatism. I consider it pragmatic to use...
Eric J. Bowman
eric@...
Nov 10, 2006 8:51 am
... matter of preference and not one of pragmatism. We are going in circles. If it doesn't matter which you choose, then in order to get consistancy I'm...
... It depends on the application. But I mostly use the approach used in SQL which is to be singular. But then again, that's hardly a convention at all. Lot's...
Nic>> But I mostly use the approach used in SQL which is to be singular. I guess what I'm looking for is "why?" Just because in SQL? -Mike Schinkel ...
... In SQL the why is because you're really talking about each individual record. So you might select a record thus: select * from user where id = 72; clearly...
Nic>> In SQL the why is because you're really talking about each individual record. Just FYI, I'm really familiar with the reason in SQL and agree with it. ......
Nic>> That says more about other frameworks than it does about R&R I think. So, you must really be liking SOAP right about now? ;-) Nic>> It doesn't mean their...
... The bluntness of my response is typical of me, myself and I -- as was the politeness of my response. It was not rude, nor was it meant as a flame, nor was...
Eric J. Bowman
eric@...
Nov 13, 2006 10:41 am
... And I didn't reference your email specifically either. :) Numerous people replied and my comments were about the cumulative, not any one in specific (at...
Dave, The question strikes me as very abstract. What might you be assessing for? Say you are interested in scaleability. You should be able to evaluate that...
... Which seemingly resolves to a subjective assessment? Fine, identify it as such, perhaps give it a low weighting. I've seen systems pointed to with clearer...
... It's not so much a matter of subjective as it is of availability of data and what can be inferred from it. This is definitely not as simple as writing...
... Or, looking for a metric, I can hack n users now without falling over, scalability for this app means hacking n*1000 users without falling over. Now...
Dave, ... That's fine, but it's off topic. We were talking about how you can use REST to evaluate a system. What you describe above would not be an instance...