1 |
On Mon, Oct 22, 2012 at 1:35 PM, Peter Stuge <peter@×××××.se> wrote: |
2 |
> Rich Freeman wrote: |
3 |
>> I still think we're re-inventing the wheel. |
4 |
> |
5 |
> Oh I've thought so all along, but you seemed very determined to do it. |
6 |
|
7 |
Well, I wouldn't call it a complete waste. I just think there is a |
8 |
point of diminishing returns. |
9 |
|
10 |
I posted to the list asking what was holding us up. After some IRC |
11 |
discussion it came up that validation was perceived as being an issue, |
12 |
and one which could be rate-limiting. If as a result of my work |
13 |
people think that validation isn't a big deal, then that's an |
14 |
accomplishment. |
15 |
|
16 |
Testing is by its nature self-defeating - if you do a perfect job the |
17 |
tests will all pass, but chances are that a team that skips testing |
18 |
will not be perfect. |
19 |
|
20 |
I'll try to make a few more incremental steps until we're a bit |
21 |
closer. Then we can do some manual comparisons. If the header issue |
22 |
is fixed then that is a bunch of work that will go away. (The paths |
23 |
in the header were the only difference I saw, but that was based on |
24 |
some manual spot checks. If that is fixed and there are just a few |
25 |
one-off situations where there isn't a match then the hash comparison |
26 |
should show them.) |
27 |
|
28 |
Rich |