Use 'frog' in apidoc. Also fix broken dependencies.

Use 'frog' in apidoc. Also fix broken dependencies.

Committed: https://code.google.com/p/dart/source/detail?r=6870

[ngeoffray, 2012-04-23 15:50:08]

LGTM, thanks Peter

[Bob Nystrom, 2012-04-23 16:03:20]

Couple of nits and questions but otherwise LGTM.

https://chromiumcodereview.appspot.com/10174006/diff/1/dart/lib/dartdoc/clien...
File dart/lib/dartdoc/client-live-nav.dart (right):

https://chromiumcodereview.appspot.com/10174006/diff/1/dart/lib/dartdoc/clien...
dart/lib/dartdoc/client-live-nav.dart:24: // Figure out where we are.
Good catch. When I first wrote this, we weren't reliably running main() after
contentLoaded automatically.

https://chromiumcodereview.appspot.com/10174006/diff/1/dart/lib/dartdoc/dartd...
File dart/lib/dartdoc/dartdoc.dart (right):

https://chromiumcodereview.appspot.com/10174006/diff/1/dart/lib/dartdoc/dartd...
dart/lib/dartdoc/dartdoc.dart:180: print('non-zero exit code from
$compilerPath');
Nit, but please make this a full sentence:

'Unexpected on-zero exit code from $compilerPath.'

https://chromiumcodereview.appspot.com/10174006/diff/1/dart/utils/apidoc/apid...
File dart/utils/apidoc/apidoc.dart (right):

https://chromiumcodereview.appspot.com/10174006/diff/1/dart/utils/apidoc/apid...
dart/utils/apidoc/apidoc.dart:67: if (compilerPath === null) {
==

https://chromiumcodereview.appspot.com/10174006/diff/1/dart/utils/apidoc/apid...
File dart/utils/apidoc/apidoc.gyp (right):

https://chromiumcodereview.appspot.com/10174006/diff/1/dart/utils/apidoc/apid...
dart/utils/apidoc/apidoc.gyp:23: '<(PRODUCT_DIR)/api_docs/index.html',
Does it matter that all of the other generated HTML isn't listed as an output
here?

[ahe, 2012-04-24 10:31:01]

Hi Nicolas and Bob,

Thank you for taking a look.

Cheers,
Peter

https://chromiumcodereview.appspot.com/10174006/diff/1/dart/lib/dartdoc/clien...
File dart/lib/dartdoc/client-live-nav.dart (right):

https://chromiumcodereview.appspot.com/10174006/diff/1/dart/lib/dartdoc/clien...
dart/lib/dartdoc/client-live-nav.dart:24: // Figure out where we are.
On 2012/04/23 16:03:20, Bob Nystrom wrote:
> Good catch. When I first wrote this, we weren't reliably running main() after
> contentLoaded automatically.

Is it just me, or is window.on.contentLoaded pointless now. This code doesn't
work with the use of window.on.contentLoaded. Or perhaps this is a bug?

Also, how would I execute code before the document is loaded?

https://chromiumcodereview.appspot.com/10174006/diff/1/dart/lib/dartdoc/dartd...
File dart/lib/dartdoc/dartdoc.dart (left):

https://chromiumcodereview.appspot.com/10174006/diff/1/dart/lib/dartdoc/dartd...
dart/lib/dartdoc/dartdoc.dart:185: process.onError = (error) {
After talking to sgjesse about this, I restored the use of onError. This detects
errors like the executable not being found.

https://chromiumcodereview.appspot.com/10174006/diff/1/dart/lib/dartdoc/dartd...
File dart/lib/dartdoc/dartdoc.dart (right):

https://chromiumcodereview.appspot.com/10174006/diff/1/dart/lib/dartdoc/dartd...
dart/lib/dartdoc/dartdoc.dart:180: print('non-zero exit code from
$compilerPath');
On 2012/04/23 16:03:20, Bob Nystrom wrote:
> Nit, but please make this a full sentence:
> 
> 'Unexpected on-zero exit code from $compilerPath.'

Done.

https://chromiumcodereview.appspot.com/10174006/diff/1/dart/utils/apidoc/apid...
File dart/utils/apidoc/apidoc.dart (right):

https://chromiumcodereview.appspot.com/10174006/diff/1/dart/utils/apidoc/apid...
dart/utils/apidoc/apidoc.dart:67: if (compilerPath === null) {
On 2012/04/23 16:03:20, Bob Nystrom wrote:
> ==

Testing null with === is safer. Based on my personal experience, I have a mental
block that prevents me from testing null with == :-)

https://chromiumcodereview.appspot.com/10174006/diff/1/dart/utils/apidoc/apid...
File dart/utils/apidoc/apidoc.gyp (right):

https://chromiumcodereview.appspot.com/10174006/diff/1/dart/utils/apidoc/apid...
dart/utils/apidoc/apidoc.gyp:23: '<(PRODUCT_DIR)/api_docs/index.html',
On 2012/04/23 16:03:20, Bob Nystrom wrote:
> Does it matter that all of the other generated HTML isn't listed as an output
> here?

That depends. I would like to identify the last file that apidoc writes out.
Would it be OK to call docIndex() as the last thing in Dartdoc.document?

However, the file index.html is not created atomically, so if you abort the
build you may get a partial index.html with a fresh timestamp.

Perhaps a simpler (and better) alternative is to write out a timestamp file at
the end of main() in apidoc.dart.

WDYT?

[Bob Nystrom, 2012-04-24 16:38:08]

https://chromiumcodereview.appspot.com/10174006/diff/1/dart/lib/dartdoc/clien...
File dart/lib/dartdoc/client-live-nav.dart (right):

https://chromiumcodereview.appspot.com/10174006/diff/1/dart/lib/dartdoc/clien...
dart/lib/dartdoc/client-live-nav.dart:24: // Figure out where we are.
On 2012/04/24 10:31:02, ahe wrote:
> On 2012/04/23 16:03:20, Bob Nystrom wrote:
> > Good catch. When I first wrote this, we weren't reliably running main()
after
> > contentLoaded automatically.
> 
> Is it just me, or is window.on.contentLoaded pointless now.

Hmm, good point. I'll send an email asking about that.

> This code doesn't
> work with the use of window.on.contentLoaded. Or perhaps this is a bug?

I think that's intended but not useful. The event only fires when the content
load actually happens. In this case, that always occurs before you've added the
listener. We don't do any special handling that says "if the event happened
before you registered, fire it immediately upon registration".

> Also, how would I execute code before the document is loaded?

This is a Vijay question, but I believe the answer right now is that you can't.

https://chromiumcodereview.appspot.com/10174006/diff/1/dart/utils/apidoc/apid...
File dart/utils/apidoc/apidoc.dart (right):

https://chromiumcodereview.appspot.com/10174006/diff/1/dart/utils/apidoc/apid...
dart/utils/apidoc/apidoc.dart:67: if (compilerPath === null) {
On 2012/04/24 10:31:02, ahe wrote:
> On 2012/04/23 16:03:20, Bob Nystrom wrote:
> > ==
> 
> Testing null with === is safer. Based on my personal experience, I have a
mental
> block that prevents me from testing null with == :-)

Well, we're changing the semantics of "==" specifically so you don't have to
worry about === here, right? If we don't trust the language's semantics, who
will? :)

https://chromiumcodereview.appspot.com/10174006/diff/1/dart/utils/apidoc/apid...
File dart/utils/apidoc/apidoc.gyp (right):

https://chromiumcodereview.appspot.com/10174006/diff/1/dart/utils/apidoc/apid...
dart/utils/apidoc/apidoc.gyp:23: '<(PRODUCT_DIR)/api_docs/index.html',
On 2012/04/24 10:31:02, ahe wrote:
> On 2012/04/23 16:03:20, Bob Nystrom wrote:
> > Does it matter that all of the other generated HTML isn't listed as an
output
> > here?
> 
> That depends. I would like to identify the last file that apidoc writes out.
> Would it be OK to call docIndex() as the last thing in Dartdoc.document?

We could do that, but that's likely to be brittle as apidoc and dartdoc change
over time. And, actually, I think right now the last thing that will be
generated is the client-side JS files since those are done async and don't even
start until main() has returned and all of the other sync operations are done.

I'm a gyp noob. What is this used for? Is it just to tell when this build step
is dirty and needs to be re-run?

> Perhaps a simpler (and better) alternative is to write out a timestamp file at
> the end of main() in apidoc.dart.

Hmm, we could do that, but putting code in apidoc specifically to make gyp happy
feels weird to me. How about we add another little python action here that just
writes a timestamp file after the main action is done? That way all of the gyp
magic is confined to gyp.

[ahe, 2012-04-25 08:40:38]

https://chromiumcodereview.appspot.com/10174006/diff/1/dart/utils/apidoc/apid...
File dart/utils/apidoc/apidoc.gyp (right):

https://chromiumcodereview.appspot.com/10174006/diff/1/dart/utils/apidoc/apid...
dart/utils/apidoc/apidoc.gyp:23: '<(PRODUCT_DIR)/api_docs/index.html',
On 2012/04/24 16:38:08, Bob Nystrom wrote:
> On 2012/04/24 10:31:02, ahe wrote:
> > On 2012/04/23 16:03:20, Bob Nystrom wrote:
> > > Does it matter that all of the other generated HTML isn't listed as an
> output
> > > here?
> > 
> > That depends. I would like to identify the last file that apidoc writes out.
> > Would it be OK to call docIndex() as the last thing in Dartdoc.document?
> 
> We could do that, but that's likely to be brittle as apidoc and dartdoc change
> over time. And, actually, I think right now the last thing that will be
> generated is the client-side JS files since those are done async and don't
even
> start until main() has returned and all of the other sync operations are done.
> 
> I'm a gyp noob. What is this used for? Is it just to tell when this build step
> is dirty and needs to be re-run?
> 
> > Perhaps a simpler (and better) alternative is to write out a timestamp file
at
> > the end of main() in apidoc.dart.
> 
> Hmm, we could do that, but putting code in apidoc specifically to make gyp
happy
> feels weird to me. How about we add another little python action here that
just
> writes a timestamp file after the main action is done? That way all of the gyp
> magic is confined to gyp.

This is not to make GYP happy. Any tool that participates in an incremental
build system needs to have certain behavior. Otherwise, you always need to wrap
the tool in another tool that runs it, and then atomically moves the result to
the right location. I know you can do that for a directory on Unix, but I'm not
sure about Windows.

Regardless, I think a tool provider needs to think about how his tool can be
used in a build system. Unlike javac and javadoc.

[ahe, 2012-04-25 08:42:45]

https://chromiumcodereview.appspot.com/10174006/diff/1/dart/utils/apidoc/apid...
File dart/utils/apidoc/apidoc.gyp (right):

https://chromiumcodereview.appspot.com/10174006/diff/1/dart/utils/apidoc/apid...
dart/utils/apidoc/apidoc.gyp:23: '<(PRODUCT_DIR)/api_docs/index.html',
On 2012/04/24 16:38:08, Bob Nystrom wrote:
> And, actually, I think right now the last thing that will be
> generated is the client-side JS files since those are done async and don't
even
> start until main() has returned and all of the other sync operations are done.

The last thing that happens is that the exit code of the compilation is checked.
However, the compilation is happening in parallel (in its own process).

[Jennifer Messerly, 2012-04-25 17:32:33]

https://chromiumcodereview.appspot.com/10174006/diff/9001/dart/lib/dartdoc/da...
File dart/lib/dartdoc/dartdoc.dart (right):

https://chromiumcodereview.appspot.com/10174006/diff/9001/dart/lib/dartdoc/da...
dart/lib/dartdoc/dartdoc.dart:99: final compilerPath = joinPaths(frogPath,
'minfrog');
FYI, there's still a reference to minfrog here. Dartdoc currently fails once it
tries to run this.

[Bob Nystrom, 2012-04-25 18:15:30]

On 2012/04/25 08:42:45, ahe wrote:
>
https://chromiumcodereview.appspot.com/10174006/diff/1/dart/utils/apidoc/apid...
> File dart/utils/apidoc/apidoc.gyp (right):
> 
>
https://chromiumcodereview.appspot.com/10174006/diff/1/dart/utils/apidoc/apid...
> dart/utils/apidoc/apidoc.gyp:23: '<(PRODUCT_DIR)/api_docs/index.html',
> On 2012/04/24 16:38:08, Bob Nystrom wrote:
> > And, actually, I think right now the last thing that will be
> > generated is the client-side JS files since those are done async and don't
> even
> > start until main() has returned and all of the other sync operations are
done.
> 
> The last thing that happens is that the exit code of the compilation is
checked.
> However, the compilation is happening in parallel (in its own process).

Also, copying over the static files is done asynchronously, so I don't think
there is any deterministic answer for what gets done last right now.