As a viewer, I generously read all six paragraphs.
And I learned nothing about the product.
~~~
If you're going to spend six paragraphs making claims, which is in my opinion a dangerous thing to do[1], you better have the user saying "yeah!" in their head successively louder as each paragraph goes by, because you're delaying the part where you actually tell the user about the product. But this page doesn't really say anything about the product.
[1] Making one claim after another is dangerous because it gives you a chance to say something stupid or disagreeable. For instance their claim:
> I’ll wager that Facebookers spend 50% of their times on tests, time which could – in an ideal world – be spent on product.
Serves to make them look a bit ridiculous. If the product is a polished app then spending time on tests is spending time on the product. It seems silly to suggest otherwise.
Stop claiming things and tell me what your damn product does. If I have to look it up what it does I'll probably find a list of competitors alongside it.
Good point. We didn't want to make this a sales pitch - we wanted to make it a call to action about the state of testing. But I can see how pitching the product would help. So here goes:
- incredibly easy to set up, you'll probably have your tests running in one click
- very fast - our users report tests run faster than on their own machines
- Supports parallelism - you can reduce your test times by up to about 7x.
Mmm, a title tailor-made for the HN frontpage but disappointingly lacking in content.
It's just a thin layer on top of a call to action for this new CI service. The service itself may be awesome but I didn't learn anything about it from this.
> Testing doesn’t really help our customers. At best it helps indirectly, in the same way that HR or payroll or marketing does.
I agree with this, but I think this applies to source code in general, not just tests. Software is different from many other crafts in that there's a strong separation between construct and artifact. The metaphor would be that the house you live in is actually an artifact/result of a different house somewhere else, that you might or might not have access to, and changes to the house you live in requires change to the "source" house. And the source house looks nothing like the actual house. As Rich Hickey puts it, users of your software rarely gets to look at the source code and be satisfied with how pleasant it is (or isn't).
This is step 1: a vastly easier-to-use and faster continuous integration server than you currently have (if you have one). The manifesto talks more about where we're going - we want to be the dev tools for your company.
9 comments
[ 4.4 ms ] story [ 27.1 ms ] threadAnd I learned nothing about the product.
~~~
If you're going to spend six paragraphs making claims, which is in my opinion a dangerous thing to do[1], you better have the user saying "yeah!" in their head successively louder as each paragraph goes by, because you're delaying the part where you actually tell the user about the product. But this page doesn't really say anything about the product.
[1] Making one claim after another is dangerous because it gives you a chance to say something stupid or disagreeable. For instance their claim:
> I’ll wager that Facebookers spend 50% of their times on tests, time which could – in an ideal world – be spent on product.
Serves to make them look a bit ridiculous. If the product is a polished app then spending time on tests is spending time on the product. It seems silly to suggest otherwise.
Stop claiming things and tell me what your damn product does. If I have to look it up what it does I'll probably find a list of competitors alongside it.
(edit: which is exactly what wikipedia shows http://en.wikipedia.org/wiki/Continuous_integration#Software)
- incredibly easy to set up, you'll probably have your tests running in one click
- very fast - our users report tests run faster than on their own machines
- Supports parallelism - you can reduce your test times by up to about 7x.
It's just a thin layer on top of a call to action for this new CI service. The service itself may be awesome but I didn't learn anything about it from this.
I agree with this, but I think this applies to source code in general, not just tests. Software is different from many other crafts in that there's a strong separation between construct and artifact. The metaphor would be that the house you live in is actually an artifact/result of a different house somewhere else, that you might or might not have access to, and changes to the house you live in requires change to the "source" house. And the source house looks nothing like the actual house. As Rich Hickey puts it, users of your software rarely gets to look at the source code and be satisfied with how pleasant it is (or isn't).
Still not sure exactly how they are improving testing. As far as I can tell, it's just running tests on their machines instead of mine.
It integrates with GitHub, and they've done something-or-other to facilitate parallelism. None of this tells me why it might be a useful service.