Google

[Ryan Mattison]

For being geniuses, your documentation and organizational skills are
less than desirable. You choose to take on multiple IDEs/environments
and release them untested. Try picking one, we all have Linux, Mac,
Windows. State the person that wrote this documentation is on XXYY
version xy, so it must work as he expected (we are testing other
environments, but we know this one works!). I've developed against a
load of sdks web, mobile, data, os, windows etc etc, mostly because I
like to have an extremely good time,can't pull my life together, need
money and lack the overall skills to be a engineer. One thing is
certain though, I would rather jab my eyes out then work through the
next Google project. I have to be employed in America, why do you
have to make it painful. Not everyone gets off on digging through
scripts, open source code, and logs to decipher why this pos decided
to lock x file and not rebuild it & write nothing to a log etc. Some
of us want to build something cool and go to the bar.

Lastly, why the verbosity? Clear & concise patterns & code never seem
to be the aim. Its not like the code needs to be clear it'll take a
pick ax to get through a relatively new Google environment anyway.
You may as well finish it off and let developers type less.

Composite, C
Calendar, Ca

Let me imagine I got the chemistry degree I should have.

Alright, better now .. back to jabbing my eyes out. Your responding/
discussion skills are way above bar though, so that is good !~

[Jeff Schnitzer]

Wow. I've found the GAE docs to be pretty nice. What some pain? Try
the docs for:

* Amazon Elastic Beanstalk
* Paypal
* Google Checkout
* Amazon Payments
* require.js

...just to name a few that have given me grief lately. Paypal gets
the poo-colored star here. They are truly a testament to how hard it
is to unseat a market-leading incumbent, even when they suck.

Jeff

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

[stevep]

Agreed. I'd like to see more from Google about architectural
considerations, but compared to FBook Credit docs G's are heavenly.

stevep

[Robert Kluin]

I find the App Engine docs pretty good too. Some things could
definitely use improving, like documenting all the rates, limits,
quotas, deadlines, etc..., in one spot so we can search / find info
(including low-level stuff), but they are otherwise ok.

I also find that if you stay *high* in the SDK files (ie the public
interface) it is not too bad -- and the doc strings are good. Just
don't dig in; it gets tangled up fast.

The gdata docs / api are a different story -- complete and total mess,
along with a disaster of an API to go with it! But, that's not
related to App Engine.

Robert

[Tim Hoffman]

I would strongly agree with Robert, and not agree with the OP.

I find the appengine docs are fine, everything can do with improvement though ;-)

Just my 2c worth

T

[Ryan Mattison]

I should have been clearer. The AppEngine documents themselves are
above par, it took only a few hours to proof of concept what I wanted
from them. Should people switch to GWT, AppEngine, GData.. Is it
going to save time, money, be more reliable. I like finding out my
making something small.

From what I've seen so far, which isn't enough time to form a quality
opinion - the GAE,GWT, and GDATA teams think it is a race, and they're
leaving pitfalls for each other's users scattered throughout.

At about hour 12-20 of prototyping you hit a cyclone in the middle.
This thing spun me so completely I thought I was going a need a team
from M.I.T. to come in to oAuth2. Something that is done & done, I
wrote a oAuth2/data library for my classic TI83+ that prototypes
better. There is a large mixture of version 2.x.x, 1.0.x, 1.0.x,
550.x.x.x (client or server side). When combining gdata,gwt,gae.
There are code branches linked with aged, dead solutions to the
simplest problems, 25 different OAuth2 pages stating/implementing the
same thing completely differently. I know the concepts, it'd be
awesome to have a functioning/reusable syntax. I must have missed the
super important golden link that has quality information (flow
charts,spreadsheets,functional repository code samples). I find a
small drawing worth more than a 40,000 word dialog of a person
mentally masturbating to his architecture.

Ketamine trips have less holes to fall in.

[Ryan Mattison]

Wrote a long winded reply not sure if it went through ->>> but this -
>>>> is hell.

The gdata docs / api are a different story -- complete and total
mess,
along with a disaster of an API to go with it! But, that's not
related to App Engine.

On Sep 21, 12:59 am, Tim Hoffman <zutes...@gmail.com> wrote:

[Mark Bucciarelli]

On Mon, Sep 19, 2011 at 11:48 PM, Ryan Mattison <rmatt...@gmail.com> wrote:

Lastly, why the verbosity?  Clear & concise patterns & code never seem
to be the aim.   Its not like the code needs to be clear it'll take a
pick ax to get through a relatively new Google environment anyway.

It's refreshing to read Tornado's code and docs after fighting

app engine.  i guess those hackers helped (a lot?) with gmail

and gmap?

query times linear with sizeof(result_set) was really sexy, 

but 2 QPS per instance?  really?  and i hit soft limit w/ 

python so often it's a joke---i can't do any serious processing

without hearing the google cash register ring away.

Simple python prevayler setup with tornado + nginx front end

pumps 7k inserts/second (one big gzip'd post).  bonus: it's 

single-threaded (remember medusa?).  since it uses prevayler,

it will be easy to replicate/coordinate tornado nodes (if i ever

need to).

disappointed too and rewriting as fast as i can,

m

[Jeff Schnitzer]

On Thu, Sep 22, 2011 at 12:10 PM, Mark Bucciarelli <mkb...@gmail.com> wrote:
>
> It's refreshing to read Tornado's code and docs after fighting
> app engine.  i guess those hackers helped (a lot?) with gmail
> and gmap?
> query times linear with sizeof(result_set) was really sexy,
> but 2 QPS per instance?  really?  and i hit soft limit w/
> python so often it's a joke---i can't do any serious processing
> without hearing the google cash register ring away.
> Simple python prevayler setup with tornado + nginx front end
> pumps 7k inserts/second (one big gzip'd post).  bonus: it's
> single-threaded (remember medusa?).  since it uses prevayler,
> it will be easy to replicate/coordinate tornado nodes (if i ever
> need to).
> disappointed too and rewriting as fast as i can,
> m

While I'm a big fan of async systems like Node and Tornado, this is
really an apples/oranges comparison. The programming models are quite
different and each strengths and weaknesses. The domain of problems
appropriate to solve with Prevayler and the domain of problems
appropriate to solve with GAE's datastore have very little overlap.

If a Tornado/Prevaylor works for you, by all means use it. But
recognize that it's a niche.

Jeff