1 |
On Wed, 22 Feb 2012 09:26:01 -0800 |
2 |
Grant <emailgrant@×××××.com> wrote: |
3 |
|
4 |
> >> Likewise for debugging the kdepim mess when it was already working |
5 |
> >> well in the old kde. What ARE they thinking? Reminds me of our US |
6 |
> >> Congress, all advertising and no product. |
7 |
> > |
8 |
> > kdepim devs are just making a simple classic mistake that's |
9 |
> > been made over and over and over again. Developers do not learn from |
10 |
> > history, every time this mistake is made the team doing it thinks |
11 |
> > *they* will be different. |
12 |
> |
13 |
> What is that classic mistake? Is it the shark jumping thing? |
14 |
|
15 |
No, the mistake is the mistakes that are always made on the second |
16 |
big project. You'd have to read "The Mythical ManMonth" to truly do it |
17 |
justice (it's a really good book for developers btw), but in a nutshell |
18 |
it goes like this: |
19 |
|
20 |
For your first big project, you will proceed very slowly and carefully |
21 |
and not take on too much, as you know you know nothing. You will |
22 |
probably make a project that does one thing and does a decent job of it. |
23 |
|
24 |
Enter the second project. Buoyed by the success of the first, most devs |
25 |
will try and build something that is waaaaaaaaaay beyond their |
26 |
capabilities - I mean, how hard can it be right? It will over-reach, be |
27 |
unbuildable and timeframe estimates will be bat-shit crazy insane. |
28 |
|
29 |
The attrition rate of second big projects is rather large. |
30 |
|
31 |
Enter the third project. Humbled by the experience of the second and |
32 |
still feeling quietly (and realistically) confident by the first, most |
33 |
devs will settle down to something useful, of wide scope and still |
34 |
achievable. |
35 |
|
36 |
This same rule seems to apply to almost every project a bunch of humans |
37 |
could tackle. |
38 |
|
39 |
So back to KDEPIM. The state of that project, the amount of bugs it |
40 |
has, the attitude of the devs, the state of the migrator scripts, the |
41 |
way Akonadi can suddenly go nuclear on you and eat your kittens, the |
42 |
mysterious disappearing mails from imap stores, and more, all of these |
43 |
things point straight to second project syndrome. The devs bit off way |
44 |
more than they can chew and now it's biting them back. |
45 |
|
46 |
But it can be fixed. All it needs is a smart leader with the balls to |
47 |
nuke all the crap code, put up with the inevitable whinging, and get |
48 |
the project back on track with a set of realistic goals. This might |
49 |
well mean chucking all of Akonadi away and admitting to themselves that |
50 |
they must stick with lots of flat files for a while longer. |
51 |
|
52 |
> |
53 |
> - Grant |
54 |
> |
55 |
> |
56 |
> > This is their second big project - the most dangerous one a dev will |
57 |
> > ever work on. Frederick P. Brooks has it all covered since the 70s. |
58 |
> |
59 |
|
60 |
|
61 |
|
62 |
-- |
63 |
Alan McKinnnon |
64 |
alan.mckinnon@×××××.com |