1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
|
Meeting with: Steven, Thorsten, Nico
Problem: execute my code only if somebody else's changed something.
Proposals:
1) Introduce new binary (Steven)
gencode:
if has_changed __file/etc/nginx.conf; then
echo mycode
fi
2) Use -f plus new variable to cover .cdist/has_code:
if [ -f $global/objects/__file/etc/nginx.conf/$has_code ]; then
echo mycode
fi
And $has_code resolves to .cdist/has_code to cover the .cdist
directory.
3) Create a new tree that contains changed objects
if [ -e $global/changed/__file/etc/nginx.conf ]; then
echo mycode
fi
4) Use a \n seperated list
if grep -q __file/etc/nginx.conf $global/changed; then
echo mycode
fi
X) General problem (add to FAQ):
When having "do something only if somebody else did something"
it is possible that users get confused, because code will not
be executed.
This may especially happen in code development:
a) User creates __file/etc/nginx.conf
b) User creates a new type __nginx that restarts
nginx, only if __file/etc/nginx.conf changed.
But __file/etc/nginx.conf has already been deployed
and thus will never be restarted.
--------------------------------------------------------------------------------
Introduce exclude/conflicts attribute for types?
Handling in cdist vs. handling in the types themselves?
Examples: __file, __link, __directory
Need to handle errors on low level for those anyway.
Conclusion: Wait until we need exclude in a different type.
Pro:
Throw error as soon as possible without touching the target.
Contra:
No needod for most cases.
Very controversal for stuff like __package and __package_gem
for instance: rails; Should those two conflict or not?
|
