Create a URI class in utils/ (based on client/util/).

Create a URI class in utils/ (based on client/util/).

Committed: https://code.google.com/p/dart/source/detail?r=3722

[ahe, 2012-01-29 14:38:00]

This is not fully ready to submit, but I wanted John to get a chance to look at
it in case he reads email Sunday.

I'll move the test methods to a proper test before submitting.

https://chromiumcodereview.appspot.com/9295043/diff/1/dart/utils/uri/uri.dart
File dart/utils/uri/uri.dart (right):

https://chromiumcodereview.appspot.com/9295043/diff/1/dart/utils/uri/uri.dart...
dart/utils/uri/uri.dart:205: main() {
I'll move this method and testUri to a test case before submitting.

[ngeoffray, 2012-01-30 13:26:35]

LGTM, with your mentioned change on the test.

[ahe, 2012-01-30 17:15:10]

Hi Bill,

Please take a look at the test suite changes.

Nicolas: I have made the diff easier to read.

[Bill Hesse, 2012-01-30 18:10:30]

One of my tasks this quarter is to make a URI library, with a Path class and
path operations, and this library really goes against my design, which is
implemented on a string, and only splits up the URI into components when they
are requested.

I'd like a chance to look at this, and see if it is really the URI library we
want to end up with.


On 2012/01/30 17:15:10, ahe wrote:
> Hi Bill,
> 
> Please take a look at the test suite changes.
> 
> Nicolas: I have made the diff easier to read.

[ahe, 2012-01-30 18:43:45]

Bill,

I'd welcome you taking ownership of this class. But I need the resolve method
right now.

The classes in utils are not part of the "official" platform API, so I don't
think it'll hinder your future work.

In fact I think this URI class will help you identify future users, and epsilon
better than me simply copying the bits from client/util to frog/leg/util. This
way you know where to shop for customers of your new URI class.

Cheers,
Peter

On 2012/01/30 18:10:30, Bill Hesse wrote:
> One of my tasks this quarter is to make a URI library, with a Path class and
> path operations, and this library really goes against my design, which is
> implemented on a string, and only splits up the URI into components when they
> are requested.
> 
> I'd like a chance to look at this, and see if it is really the URI library we
> want to end up with.
> 
> 
> On 2012/01/30 17:15:10, ahe wrote:
> > Hi Bill,
> > 
> > Please take a look at the test suite changes.
> > 
> > Nicolas: I have made the diff easier to read.

[Jennifer Messerly, 2012-01-30 18:44:55]

On 2012/01/30 18:43:45, ahe wrote:
> Bill,
> 
> I'd welcome you taking ownership of this class. But I need the resolve method
> right now.
> 
> The classes in utils are not part of the "official" platform API, so I don't
> think it'll hinder your future work.
> 
> In fact I think this URI class will help you identify future users, and
epsilon
> better than me simply copying the bits from client/util to frog/leg/util. This
> way you know where to shop for customers of your new URI class.
> 
> Cheers,
> Peter
> 
> On 2012/01/30 18:10:30, Bill Hesse wrote:
> > One of my tasks this quarter is to make a URI library, with a Path class and
> > path operations, and this library really goes against my design, which is
> > implemented on a string, and only splits up the URI into components when
they
> > are requested.
> > 
> > I'd like a chance to look at this, and see if it is really the URI library
we
> > want to end up with.
> > 
> > 
> > On 2012/01/30 17:15:10, ahe wrote:
> > > Hi Bill,
> > > 
> > > Please take a look at the test suite changes.
> > > 
> > > Nicolas: I have made the diff easier to read.

lgtm too

[ahe, 2012-01-30 18:56:41]

Bill,

Just to be clear: I don't plan to submit this until I have had a chance to talk
to you in person tomorrow.

Cheers,
Peter

On 2012/01/30 18:43:45, ahe wrote:
> Bill,
> 
> I'd welcome you taking ownership of this class. But I need the resolve method
> right now.
> 
> The classes in utils are not part of the "official" platform API, so I don't
> think it'll hinder your future work.
> 
> In fact I think this URI class will help you identify future users, and
epsilon
> better than me simply copying the bits from client/util to frog/leg/util. This
> way you know where to shop for customers of your new URI class.
> 
> Cheers,
> Peter
> 
> On 2012/01/30 18:10:30, Bill Hesse wrote:
> > One of my tasks this quarter is to make a URI library, with a Path class and
> > path operations, and this library really goes against my design, which is
> > implemented on a string, and only splits up the URI into components when
they
> > are requested.
> > 
> > I'd like a chance to look at this, and see if it is really the URI library
we
> > want to end up with.
> > 
> > 
> > On 2012/01/30 17:15:10, ahe wrote:
> > > Hi Bill,
> > > 
> > > Please take a look at the test suite changes.
> > > 
> > > Nicolas: I have made the diff easier to read.

[Bill Hesse, 2012-01-31 10:31:18]

LGTM.

https://chromiumcodereview.appspot.com/9295043/diff/5008/dart/tools/testing/d...
File dart/tools/testing/dart/test_runner.dart (right):

https://chromiumcodereview.appspot.com/9295043/diff/5008/dart/tools/testing/d...
dart/tools/testing/dart/test_runner.dart:82: "
${configuration['mode']}_${configuration['arch']}";
Does this not fit on one line if you use ${configuration['component'} ?
If you need to split it, then you should uniformly use + or uniformly use $. 
Also, could the first part go on the previous line?

[Bill Hesse, 2012-01-31 10:37:56]

https://chromiumcodereview.appspot.com/9295043/diff/1/dart/utils/uri/uri.dart
File dart/utils/uri/uri.dart (right):

https://chromiumcodereview.appspot.com/9295043/diff/1/dart/utils/uri/uri.dart...
dart/utils/uri/uri.dart:278: base.resolve("g#s/../x").toString());
Should you also test parameters in the path elements?
resolve("../g;p/h;s")?

[ahe, 2012-01-31 11:43:00]

I forgot to "git add" the new test and test suite. PTAL.

https://chromiumcodereview.appspot.com/9295043/diff/1/dart/utils/uri/uri.dart
File dart/utils/uri/uri.dart (right):

https://chromiumcodereview.appspot.com/9295043/diff/1/dart/utils/uri/uri.dart...
dart/utils/uri/uri.dart:278: base.resolve("g#s/../x").toString());
On 2012/01/31 10:37:57, Bill Hesse wrote:
> Should you also test parameters in the path elements?
> resolve("../g;p/h;s")?

That's being tested on lines:

243-246
269-270

But your particular pattern isn't tested (leading ..) and I think it makes sense
to add it.

https://chromiumcodereview.appspot.com/9295043/diff/5008/dart/tools/testing/d...
File dart/tools/testing/dart/test_runner.dart (right):

https://chromiumcodereview.appspot.com/9295043/diff/5008/dart/tools/testing/d...
dart/tools/testing/dart/test_runner.dart:82: "
${configuration['mode']}_${configuration['arch']}";
On 2012/01/31 10:31:19, Bill Hesse wrote:
> Does this not fit on one line if you use ${configuration['component'} ?
> If you need to split it, then you should uniformly use + or uniformly use $. 
> Also, could the first part go on the previous line?

Done.

[ngeoffray, 2012-01-31 11:46:30]

LGTM

[Bill Hesse, 2012-01-31 11:46:56]

LGTM.

[ahe, 2012-01-31 12:11:59]

Hi John, Bill, and Nicolas,

Thank you for taking a look, and sorry for making it so confusing :-)

Cheers,
Peter