Exclusive suites,specs

[wclr]

There is the issue in Pivotal Jasmine repo that well describes the problem:
jasmine/jasmine#181 (comment)

what do you think is it worth to implement?

[tj]

i dont get it, what is "iit" etc?

[domenic]

This is actually really useful. It's like the opposite of xit, it says "if I have iit-ed a given test, run it and no others". In other word it xits all other tests for you.

Yeah, grep works too, but sometimes iit is just a bit easier.

[tj]

oh i see, nice feature, weird implementation haha "xit" is weird too though IMO

[paulmillr]

sounds like crap for me. 👎

[Raynos]

This is a very good feature.

The problem is that it's hard to name.

Being able to run individual describe or it blocks by editing the code rather then using --grep is useful

[tj]

i dig it, i was sceptical about xit etc but the implementations are small which is more what matters to me. I dont want a ton of extra features if they add considerable weight, but I think we could easily translate these into --greps and reuse that logic

[tj]

for example iit (with a better name? haha) could append @run to the title, and grep that, the rest just works

[Raynos]

@visionmedia if you implement iit we also need ddescribe and ccontext

[Raynos]

The better API might be

it(name, function, true)

where the optional last boolean turns it into an "only run this block"

[tj]

yeah the first char thing is kinda silly, I like that it's super convenient, "xit" at least sort of reads like what it does. I dont like context though I want to remove that, it and describe are acceptable globals but not so much context

[tj]

same with specify etc, those should be a different UI if anything

[Raynos]

@visionmedia if you implement this either kill the context global or allow this pattern to work for all globals including test and suite and other interfaces

[tj]

yeah this pattern would have to be for the others as well. The globals don't bother me though, xdescribe etc are completely irrelevant

[wclr]

Instead of xdescribe, ddescrbe, xit, iit maybe it would be better to use symbols at the beginning of the name of test, for example # for pending and @ for exclusive:

describe('#should be pending test and should not be run', ... )

and for exclusive tests use @ symbol at the beginneing of the name:

describe('@should be exclusive test to run', ... )

[Raynos]

That breaks backwards compatibility

[wclr]

@Raynos You mean for someone who used that symbols at the beginning for the test names? Well, I think if this method is more convenient for a workflow difficulty with backwards compatibility could be overcome.

[tj]

@whitecolor yeah I'd prefer something like that personally, it would automatically work for TDD etc too, and exports can't support this feature unless we use the titles +1 from me that's a good alternative, we can always deprecate "xit" and let that sit for a few releases

[tj]

not sure about those chars though neither of them scream exclusive or pending to me

[wclr]

"!" - could be used for pending
"*" - for exclusive

or ("?" for pending and "!" for exclusive)

some symbols could be chosen as defaults but also made configurable via parameters and via mocha setup in browser.

[domenic]

Guys just copy ddescribe and iit and be done with it. Or at least do that as well. It's simple, intuitive, and compatible with other BDD interfaces.

[tj]

I definitely wouldn't call it intuitive, I had no clue what it was from looking at it, chars wouldn't be any better though. only_describe( etc would be better IMO, 4 more chars I guess :p but at least it reads ok

[tj]

definitely want this lately, anyone hate only_ as a prefix? iit just looks like a typo

[rauchg]

I vote for some sort of title prefix, maybe !!

[wclr]

only_ seems to be too elablorated. But why not make two options for a developer: either use "only_" or shorter variant "o" (odescribe, oit to be consistent with xdescribe, xit syntax).

[tj]

too elaborate? because it's clearer? I'm not a fan of xit either personally, I'd rather have ignore_it or similar, it reads better, more obvious at a glance, less easy to accidentally leave in your source

[rauchg]

xit I didn't immediately get personally, and iit just put me in limbo

[wclr]

Personally I like only_ and ignore_ because its clear, right. But I think its good to have shortcuts (like "x" and "o" prefix) either. I vote for only_ and _ignore and "o" and "x" shortcuts.

[rauchg]

I'm gonna lobby for it.skip( it.only( tomorrow

[domenic]

I like it.skip and it.only, but I would again encourage iit and ddescribe additionally, just for equivalence with existing libraries (JSTestDriver/Testacular/Angular.js's Jasmine fork, mainly).

I also really want a shorthand. Like xit and xdescribe, these will be things I quickly add and never commit to source control, so clarity is not important to me at all, but taking one keypress to insert definitely is. I want to add a single character, press Ctrl+S, and have the watch feature automatically run that test and that test only.

[tj]

-1 on shortcuts for me, if you're using these that often you have bigger issues IMO, it doesn't need to be that convenient . I like your suggestion @guille, less globals, a little more annoying to jump a word and type .skip but haha that goes back to the who cares thing

[szkrd]

iit and ddescribe is a better way to catch these keywords with lint, so while you could develop locally and ignore lint errors, the ci build tool would catch them. Is this clear?

I also find it weird that xdescribe and xit was fine, but ddescribe and iit wasn't
#263

Would it be possible to reconsider this decision?