Slashdot Mirror
Are Googlers Too Smart For Their Own Good?
theodp writes "If you're a mere mortal, don't be surprised if your first reaction to Google Storage for Developers is 'WTF?!' Offering the kind of 'user-friendly' API one might expect from a bunch of computer science Ph.D.s, Google Storage even manages to overcomplicate the simple act of copying files. Which raises the question: Are Googlers with 'world-class programming skills' capable of producing straightforward, simple-to-use programming interfaces for ordinary humans?"
So, theodp, if you were a developer you would look at this and see a set of interfaces to web services done in a RESTful manner. You would say, "Oh, my users want to use Google storage but they need more of a drag and drop interface." Then you would spend a couple weeks using Ruby on Rails and Scriptaculous to make virtual folders or buckets or whatever your application calls them and using the elegance of RoR with the UI of Scriptaculous so the user can move their photos or data from your server to the cloud or vice versa. You could really use anything you want to interact with it but I would bet these two GPL compatible tools would result in the most rapid of web application development.
So three sentences with links to Google besmirching them for being smart will get you on the frontpage of Slashdot these days? Really the substance of the 'story' here is essentially "WTF?! So complicated it must Suck!"
Offering the kind of 'user-friendly' API ...
Here's a final hint: API stands for Application Programming Interface is not supposed to be user-friendly. It's supposed to be developer-friendly. I hope I don't sound like a Google fanboy but this is a nontrivial task and I would defend the API they have produced. The documentation is far more than you would get from a CS PhD. You want me to take notice of your mindless drivel, theodp? Get off your ass, code an interface for this API and then point out how the API and documentation is lacking in a step by step post. That would be helpful and deserve a place in Slashdot's programming section. What you have here is not.
My work here is dung.
They wouldn't call it code.
I still cannot find the droids I am looking for...
In other news: the space shuttle UI is too complicated for regular car drivers! duh.
Whatever happened to simple interfaces, like:
"Would you like to play Global Thermonuclear War? [YES|NO]"
Seven puppies were harmed during the making of this post.
We didn't have a Google story for over two hours, so we had to post what was available.
The only nonintuitive thing is the name "bucket", which might be better called "zone" or "filesystem". Other than that, it looks like it provides just about what I'd expect of a high-level filesystem representation.
Sheesh, just think about what the complaints would be if they provided something closer to VFS-type mappings so people ended up commonly rewriting half of FUSE to get their data where they like.
For every problem, there is at least one solution that is simple, neat, and wrong.
This isn't to discredit the idea of ease of use or good design - god knows Google graphs requires way more hoops than it should (compare, say, Visifire).
I think it's easy to look at the developer's guide and just flee in terror, but honestly if that's your reaction, Google storage API is probably not the droid you're looking for. If you need simple file sharing that a typical user can appreciate without having to read a manual, Dropbox may be more appropriate; Google Storage API is written with developers in mind.. I'm a big fan of some of Google's APIs, Dropbox, and Google Docs for sure.
UI != API. A proper API doesn't need to be simple, it needs to work properly and consistently. Not to get too subtle, but a complex API for something like this is perfectly fine as long as it is not overly complicated, if you get what I mean.
I still cannot find the droids I am looking for...
I was expecting something really crazy and complex but what I saw was well documented and made sense. Seriously, how on earth is this front page news on slashdot?? I wont repeat the many well made statements that "API's arent for users" above. I'm just surprised this has made it to the front page as a developers link. I sure hope I don't work with the sub. at any point if he thinks this is an example of people being "too smart for their own good". /saddened
jaymz
As the documents point out, it's the same API used for Amazon EC3 and others. They're implementing someone else's protocol.
who flood developer-boards with questions that typically look like
" Sir Sir please help sir I have project due sir I need full workking code by tomorrow sir" ??
If so, you would expect everything to be point and click, I guess.
Just a quick FYI, API does not mean UI. I noticed some of the slashdotters were conflating the two.
"Anything tastes good if you deep fry it."
You could just set your filter for "mere mortal" appropriately and you won't see these things anymore.
Oh, yeah, it's not easy to pad these out to 120 characters.
There's been an awful lot of discussion about what is or isn't simple, and people have gotten a pretty sophisticated notion of simplicity, but I'm not sure it has helped.
-- Ward Cunningham
Colorless green Cthulhu waits dreaming furiously.
I think this is a terrible idea. It violates the principle of "make it as simple as possible, but not any simpler". Some things are just complicated. Even if the UI is nice and clean, what it interacts with is not and the users have to know about that. I don't think any UI could be understood by a 5 year old and I'm fine with that.
What exactly is hard to use or cryptic about a RESTful API? If such a thing strains your brain too much you probably are in the wrong line of work. I'd recommend you get a job flipping burgers but even that may be way more than your intellectually capable of.
Things become much more complicated then first impression when you try to really explain something. For example I went to a football game with a group of Chinese grad students and they asked me how a team can score points. I thought to myself this is easy, and began to explain the rules.
1. Touchdowns are worth 7 points... err they are worth 6 points technically
2. After a touchdown the scoring team can decide to kick the ball through the uprights for 1 point
Or
3. The scoring team can decide to run another regular play and if they enter the end-zone again on that 1 play they get 2 points.
4. Fields goals are 3 points and are scored when the team on offense can kick the ball through the uprights.
5. The defense can score points if they can tackle an offensive player in the end-zone while they are holding the football. The defensive team then gets 2 points and gets the ball kicked to them on the following play instead of the normal system where the scoring team kicks the ball to the other team.
6. If the defense can steal the ball and run into the end-zone they are facing then it is a touchdown and rule 2 and 3 apply.
By the end of this discussion they were more confused then when we started. So when you say how hard can it be to explain how to store a file questions like.
1. How to delete?
2. How to rename?
3. How to create folders or other organizational structures?
4. How to move items between organizational structures?
5. How to copy an item already in storage?
6. How to download multiple files?
7. Can security be set or changed?
8. Oh yeah and how to I upload a file in the first place?
The more precision you apply to a discussion the more complicated they tend to get. Just like a touchdown is 7 points is easier to understand, upload a file is easy too.
As the parent to your post noted: we are talking about an API here. Precisely none of it is user facing.
After skimming the file-copying code, I agree with the people who say it's not complicated. I'm not a Python programmer either. The example functions they gave look like good starting points for wrappers that would provide the higher level, "get, send, delete" sort of functionality the poster wants. The only thing that confuses me is why you have to have "config = boto.config" when the config variable isn't used in the rest of the code. To me, it looks like you're only interested in the side effects of retrieving the configuration and not the result. Couldn't you just "boto.config()" or something at program startup? Of course that's probably more of a Python question from somebody who is ony passably familiar with the language. It's nothing complicated about the API.
For all intensive purposes, "whom" is no longer a word. That begs the question, "who cares"?
Yes you are. This is not a "storage system to be used as a filesystem" it's an implementation of the Amazon S3 interface that provides remote, redundant key/value storage (where the value in this case is a bucket of bytes). There's nothing to stop you implementing a file system on top of it; but the API provided by Google is at a lower level than that. Which is a good thing as a standard file system is not necessarily the best way to use this kind of storage.
The tag on the article "submittertoostupid" pretty much says it all here folks.
Got Code?
“Debugging is twice as hard as writing the code in the first place. Therefore, if you write the code as cleverly as possible, you are, by definition, not smart enough to debug it.”
- Brian Kernighan
It seems Googler's may be smart enough for their own good, but not smart enough to debug the cloud
This looks like a nice low-level API for doing really interesting and complicated things. Unfortunately, they neglected to include a high-level API to deal with what will be by far the most common use cases. Sure, it's not so difficult to implement an upload_file(filepointer, uri) function with this, but given the huge proportion of developers using this library that are going to need exactly this sort of function, do we really need all of them reinventing the wheel?
Powerful and complex functionality is good, but the most common use cases got that way for a reason. Specifically accounting for them, even if only through a set of basic frontend functions, brings major productivity boosts to the programmers that use your library. It is a thing worth doing, and it sounds like the Google folks neglected to do that in this case.
ALL of it is user facing. That's the very point of an API. The user is the developer.
This is a very, very important concept. As I said in my other post, this is a good API, a usable API. But so many APIs aren't usable. API usability should ALWAYS be considered when releasing a public library/service.
This is about developers. http://www.youtube.com/watch?v=8zEQhhaJsU4
Are you just simply way too “dumb”* for the 21st-fuckin-century?
I know I’m (sadly) a minority here. And I know that I will probably get modded to into oblivion. But except from the stupid overengineering... come on!
How about for a change actually learning something, when it is useful for you?
* I’m not even really saying that people are too dumb. It’s just that most people grew up in a culture, where it made more sense, to complain and feel entitled, to getting spoon-fed, than to understand it themselves. Where intelligent people get hate, and dumb people get special treatment (e.g. it not being allowed to point out that fact about their mental performance).
So naturally, they choose the more efficient way.
But the thing is, that we all are very much capable of grasping those complex concepts that we always say we were too dumb for. It’s just an excuse. And the more it is used, the more mental growth we miss. So after some time, we really have a hard time using our brains. Just like with a muscle. Just like we all are born with the ability to some day run for hours, every day, in the heat.
So, no, they are not too smart. We’re just used to being lazy as hell.
Any sufficiently advanced intelligence is indistinguishable from stupidity.
When someone tries to bash a cleanly designed RESTful interface as being "too complicated", you know it's a sad state of affairs. If this can lead to even one person reading chapter 5 of Fielding's dissertation, maybe some good can come of it after all...
http://www.ics.uci.edu/~fielding/pubs/dissertation/rest_arch_style.htm
http://en.wikipedia.org/wiki/Representational_State_Transfer
http://en.wikipedia.org/wiki/Roy_Fielding
"Developer", meet HTTP..
After reading through the API, if anything, it's too simple. You can't copy a bucket without reading it from Google's servers and writing it back, which is far slower than a copy carried out within their high-speed network. The "list" capability isn't well documented. The security model is about as dumb as the UNIX/Linux one; it doesn't have capabilities or anything like that. Bucket transactions are themselves atomic, but there are no user-specified atomic transactions. You can't, for example, rename "current" to "old" and "new" to "current" as an atomic transaction. (That's a normal operation in SQL, and a useful one when you've constructed a new copy of a mostly-static table and want to make it live.) Nor do buckets have version management. There's no way to read replication status; although bucket data is supposedly replicated, when does this happen? Right after uploading a bucket, or some time later?
theodp in this post quotes from a book entitled "The Dumbing-Down of Programming."
Not content with infantilizing the end user, the purveyors of point-and-click seem determined to infantilize the programmer as well.
Judging by this story submission, it turns out he's for it.
Your brain is not a computer.
It may seem a bit hilarious, apparently this kind of crap (like having bucket names conform to DNS) happens when you want to use web services as your OS. Not too hard if you just implement this once.
This bit is just silly, good for giggles but are they serious about requiring zone editing to expose a database table? Nooooo....
I didn't quite catch how you copy data to other domains, since it looks like you use a gs:// prefix to reach google storage but you say gs://cats and it is still in your account not at google's root server.. kind of annoying though maybe there's a way around it?
I think the 1024 byte limit is totally bogus, that's pretty short if it has to hold the URI path through your virtually nested buckets. Although I've seen Windows flake out at 255 character paths.. That and the bit about a "flat hierarchy", which is an oxymoron, and how you can't nest buckets but you can do so "virtually" by putting slashes in your bucket names, as if it isn't just a normal URI, they're just joshing you, a little bit of fun y'know. "Bare metal" indeed, more like stripping the metaphor down to bare CGI.
It is funny you have to allocate your own temporary file as a buffer for uploading a file, though of course that's what happens in Perl CGI. Which then makes you wonder why you cannot set a max upload data size for your app.. Of course the GSUtil command line tool looks pretty simple.
Otherwise, Animats' post is to right to the point. It isn't really that great. Kind of a bare minimum is more like it. And they stick with REST... so you should hope for a nearby library to exist that will save you not have to start implementing wierd HTTP verbs.. you have to really want this as implementing it seems as much fun as pulling teeth slowly.
So the half-hourly Apple story was already done? ;)
Any sufficiently advanced intelligence is indistinguishable from stupidity.
Sure. But you shouldn't be able to explain Photoshop or vi to a 5 year old, either.
Sure I can. Photoshop lets you paint on pictures, and Vi is like a piece of paper that you can write on.
With an API the difference is that you should be able to assume that your user will have a common lower bound on their knowledge. If your API deals with multi-threading, to be effective you probably need to assume your user knows the fundamentals of multi-threaded programming. Or, at least that the user has some base level of knowledge in computer science.
Attempting to over-simplify a concept to a child limits our ability to develop for things that aren't simple to begin with.
Write your representatives! Repeal the 2nd Law of Thermodynamics!
I wonder if Google Storage can be abused as a way to host phishing pages?
There's a phishing page that's been on Google Sites since February. Google is good about kicking off most phishing pages, but this one is different. Here's the phishing page as a web page. The actual hostile page (which is a bogus login page for Stickam) is on the "Click here to download your attachment". The actual url is http://2699962600425641406-a-1802744773732722657-s-sites.googlegroups.com/site/stickamcomlogindo/login.html?attachauth=ANoY7cpc6fembideFQyYULstnVDU-XMkgwzNLFkUv77Suh8bUq_LGrFRQ-RtLkw6pEPJb5Vk0XW4JMbOVQtqT_R6CjNCh5N2r29quoFkE5Cq1XQXUFhuegVtr4kQUMN9T3dT3yO1q-FthiahDl45UqMmFfD6gKSYwQP4bsgVoM-N5cQN0hHRvDZskuvmTdy0lqnQqUhmKFYP&attredirects=0. That's probably a page in Google Storage.
This raises the question of whether Google should be running hostile-code checks on publicly-accessible Google Storage pages.
Why would anyone take advice from a guy with an AOL email?
Why is it so hard to only have politicians for a few years, then have them go away?
Yes, and paper mills are consumers of raw pine, but that doesn’t necessarily mean all the rough edges need to be filed off it first.
Alexander Peter Kristopeit bought his basement from his mommy for one dollar.
If only they had to be worthy of the moniker to get a job...
Confucius say, "Find worm in apple - bad. Find half a worm - worse."
Developing good interfaces and good APIs is harder that you think. This maybe isn't Google finest code but it is not amateur hour either. They may need to bring in some "senior-Google-talent" but for first crack this is not bad.
Lay off these guys unless you can go it better.