| 1 |
-----BEGIN PGP SIGNED MESSAGE----- |
| 2 |
Hash: SHA256 |
| 3 |
|
| 4 |
On 18/01/14 01:10, Tom Wijsman wrote: |
| 5 |
> Should writing tests be a requirement? When? |
| 6 |
Thank you for raising the issue. We need a clear policy on this. (And |
| 7 |
this policy should be obviously documented.) And in my opinion, tests |
| 8 |
are worthless unless we do some variant of TDD[0]. |
| 9 |
|
| 10 |
We should write tests before we write implementations. This makes you |
| 11 |
think more deeply about your design, and it forces you to actually |
| 12 |
*write* the tests. I have seen time and time again projects where |
| 13 |
tests are supposed to be written post-implementation. Usually, tests |
| 14 |
are written for the first month (that's an optimistic estimate), and |
| 15 |
then the implementation is so far ahead that tests are simply dropped |
| 16 |
altogether. |
| 17 |
|
| 18 |
|
| 19 |
[0] <https://en.wikipedia.org/wiki/Test-driven_development>, for |
| 20 |
those of you who are unfamiliar with TDD. |
| 21 |
|
| 22 |
- -- |
| 23 |
Alexander |
| 24 |
alexander@××××××.net |
| 25 |
http://plaimi.net/~alexander |
| 26 |
-----BEGIN PGP SIGNATURE----- |
| 27 |
Version: GnuPG v2.0.22 (GNU/Linux) |
| 28 |
Comment: Using GnuPG with Thunderbird - http://www.enigmail.net/ |
| 29 |
|
| 30 |
iF4EAREIAAYFAlLZ3sAACgkQRtClrXBQc7WUUAEAni2RZNcbR8wyYKSJWAPXlgaR |
| 31 |
/CB3JDkzK3ZcOTn5sKcBAJ3tZ3jV+LLBg/Ls1f651gd6jECAfBvWiYAMFa8aUvSa |
| 32 |
=x2nL |
| 33 |
-----END PGP SIGNATURE----- |
Archives