[webkit-dev] Simplifying syntax in test_expectations.txt (bug 86691)

[Maciej Stachowiak]

On May 17, 2012, at 3:37 PM, Dirk Pranke <dpranke at chromium.org> wrote:

> On Thu, May 17, 2012 at 3:21 PM, Ryosuke Niwa <rniwa at webkit.org> wrote:
>> On Thu, May 17, 2012 at 3:01 PM, Dirk Pranke <dpranke at chromium.org> wrote:
>>> 
>>> As I said before, I believe they increase the readability of the file.
>> 
>> 
>> I see them as pure noise.
>> 
> 
> Clearly, different people can have different syntactic preferences :).
> 
>>> I believe the cost of learning to put delimiters in is near zero,
>> 
>> 
>> That clearly isn't near zero. Or else people wouldn't be complaining about
>> it. To quote Darin's response:
>> 
>>> Seriously, syntax is a significant barrier. Having to know which special
>>> characters to use. I don’t see this “clear delineation” you speak of. Just
>>> special punctuation I have to use to satisfy the computer
> 
> You don't have to re-quote this, I already did in this thread (and
> responded to it).
> 
> With all due respect to Maciej and Darin, neither of them have spent
> any significant amount of time working with test_expectations.txt
> files. While I appreciate that it's nice for the syntax to be
> approachable for newbies, I'm not inclined to bias in favor of newbies
> over people who are experienced. Of the people who have actually used
> the file, so far you're the only person who's spoken up as not liking
> them. Since different people prefer different things, I'm inclined to
> go with the majority of experienced users here. I am sorry if that
> means you lose out; I don't like it if anyone is unhappy and would
> prefer it if we could please everyone.

If you reject the input of people who are not yet users of test_expectations.txt, you probably won't get new users of text_expectations.txt. That would be bad for the project, so I hope that's not your final answer.

More generally, I think understandability (whether for news or experts) should take priority over familiarity. 

Let's take an example. "TEXT" next to a test name apparently means that the text fails. There is no way in the world I would guess that just from reading an expectations file. This is only conceivably understandable to someone who is an expert on the format. If someone used TEXT in code to mean "fail", I would r- their patch for failure to use meaningful identifiers.

Regards,
Maciej