Dave Liebreich wrote:
Cool.

Yes. But the exectests.sh script in that directory is not so cool :/

I know there are several in-progress efforts around unit testing. Which
one should I be tracking? I can't wait to delete my lame bash script.
-Rob

dev-quality mailing list
dev-quality (AT) lists (DOT) mozilla.org

Dave Liebreich wrote:
Robert Sayre wrote:

>Yes. But the exectests.sh script in that directory is not so cool :/
>
As a tiny step forward, I've got a simple patch to remove it and run
from a makefile target. Should I file a bug?

Yes, definitely.
-Rob

dev-quality mailing list
dev-quality (AT) lists (DOT) mozilla.org

Dave Liebreich wrote:
Cool.

The feed tests should really include the feeds we have in our
repository, i.e.

%3D

There are two feeds with issues, german (de) and norwegian (nn-N).

German has discoverability and encoding issues (bugs filed),
has newlines in the titles, which
mess up the livemark.

Axel

dev-quality mailing list
dev-quality (AT) lists (DOT) mozilla.org

Dave Liebreich wrote:
Axel Hecht wrote:
>Dave Liebreich wrote:
Cool.


>>
>The feed tests should really include the feeds we have in our
>repository, i.e.

I'd rephrase that:

When we find a bug in feed parsing, we should add a test case that
covers that bug.

Relying on an external feed to remain constant is a bad idea.

There are two kinds of tests, one is, feature tests. The other is, real
life tests.
And we do need the real life tests, whether that is due to the content
changing or us doesn't matter a bit in the end, we need to work on the
feeds we ship.

>>
>%3D
>>
>There are two feeds with issues, german (de) and norwegian (nn-N).
>>
>German has discoverability and encoding issues (bugs filed),
>has newlines in the titles, which
>mess up the livemark.

bug #s?

, duped to
, which is happily
waiting for regressions to pop up and didn't make a2.

This is not all, though. es-AR and he do show problems, too. Bottom
line, they're not tested. Bad.

Axel

dev-quality mailing list
dev-quality (AT) lists (DOT) mozilla.org

Axel Hecht wrote:
Dave Liebreich wrote:
>Axel Hecht wrote:
Dave Liebreich wrote:
Cool.

The feed tests should really include the feeds we have in our
repository, i.e.
>>
>I'd rephrase that:
>>
>When we find a bug in feed parsing, we should add a test case that
>covers that bug.
>>
>Relying on an external feed to remain constant is a bad idea.
>>

There are two kinds of tests, one is, feature tests. The other is, real
life tests.
And we do need the real life tests, whether that is due to the content
changing or us doesn't matter a bit in the end, we need to work on the
feeds we ship.

%3D

There are two feeds with issues, german (de) and norwegian (nn-N).

German has discoverability and encoding issues (bugs filed),
has newlines in the titles,
which mess up the livemark.
>>
>bug #s?

, duped to
, which is happily
waiting for regressions to pop up and didn't make a2.

This is not all, though. es-AR and he do show problems, too. Bottom
line, they're not tested. Bad.

Axel

Filed meta-bugish 337352

Axel

dev-quality mailing list
dev-quality (AT) lists (DOT) mozilla.org

Axel Hecht wrote:

Filed meta-bugish 337352

Livemarks don't use the toolkit code yet, and it's not planned to move
them to it until after FF2.

The only bug I see that I can unit test is the RTL bug. Right now, we're
stripping all markup, so the RTL div naturally gets dropped, too.
-Rob

dev-quality mailing list
dev-quality (AT) lists (DOT) mozilla.org