1 |
On 01/18/2010 09:53 AM, Alan McKinnon wrote: |
2 |
|
3 |
> Just as code is read many more times than it is written, so is a package |
4 |
> configured by the end user many more times than the config parser studied by |
5 |
> the developer. |
6 |
> |
7 |
> Your post makes sense until you realise that the use of XML in a configuration |
8 |
> designed to be changed by the user renders the package virtually unusable. |
9 |
> Given a choice between me as a developer struggling with a config parser |
10 |
> versus vast swathes of users dumping the package because of the same parser, |
11 |
> I'd say it's me that has to work harder, not my users. |
12 |
|
13 |
Back when git was still fairly fragile and Linus and the other git geeks were |
14 |
discussing how to construct the git config files, quoth Linus: "The X in XML |
15 |
stand for crap, and they couldn't even spell crap correctly." |
16 |
|
17 |
That's a matter of record in the git mail-list repository -- I didn't make it |
18 |
up, honest! |
19 |
|
20 |
To see what they finally decided upon, just look at the .git/config file in |
21 |
any git repository. Certainly more human-readable than XML, but whether it's |
22 |
lexically equivalent/superior/inferior I certainly don't know. I do know that |
23 |
Linus's opinions are usually quite clear :) |