-
Notifications
You must be signed in to change notification settings - Fork 188
/
HOWTO.add_another_format
106 lines (73 loc) · 3.86 KB
/
HOWTO.add_another_format
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
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
How to add another build format
===============================
To support yet another package format to be used with standalone build
script and Open Build Service you need to follow these steps:
Let's call the new format "XYZ" here.
General notes about the implementations. There are two areas:
SAFE implementations:
Some code runs outside of protected environments like KVM. Therefore
this code must be implemented with security in mind. Special crafted
build descriptions or binary files must not be able to exploit this code.
What makes it even more interessting is that this code can also run
on all kind of old or obscure systems. So any external dependency should
be avoided as well.
This means in short:
- code must be as simple as possible.
- code must not allow to execute random commands or to access random files.
- avoid external dependencies. When you look for a simple XML parser
check the kiwi support for this.
- code must stay compatible for all versions
Build code running inside of environment.
- using any tool is fine here. However, the tool must be installed
somehow into the build system. In best case via some dependency.
- Incompatible changes can be implemented inside of these packages
pulled into the build environment.
- external network access is not possible here.
1) Implement the parse() function into Build/XYZ.pm
parse() extracts the build dependecies from the build specification.
For RPM this would be the <package>.spec file for example.
Must be a SAFE implementation.
2) Add a query() function to Build/XYZ.pm
query() extracts information from built packages. In the rpm world
these are the .rpm files.
query returns a hash containing:
name, epoch, version, release, arch, description,
provides, requires, hdrmd5
hdrmd5 is some unique identifier of the package built, it might be
just a md5 over the entire file.
Must be a SAFE implementation.
3) Add a queryhdrmd5() function
this functions is a specialized version of query(), it just returns
the hdrmd5
Must be a SAFE implementation.
4) Add a verscmp() function
verscmp() compares two package version strings. For rpms, a version has
the form [epoch:]version-release
Must be a SAFE implementation.
5) Implement build-pkg-xyz functions
Those functions are used to setup the build environment. I.e. they
need to install/unpack the packages
Must be a SAFE implementation for the pre-installation part. Afterwards
it is fine to use any tool to install the packages (like rpm itself).
6) Implement build-recipe-xyz functions
This functions are called to create the build result
7) For standalone build support (actually not needed for OBS integration, but
it makes development easier) we need a "createxyzdeps" helper script.
For xyz://<...> repos it needs to download the repository metadata
and convert it to build's representation. See createyastdeps and createrepomddeps.
Special notes for non-OSS systems
=================================
Systems like MS-Windows, MacOSX or SunOS could be support as well. However, these
systems can not be installed by packages from scratch. So using preinstallimages
would be mandatory here. Support for that exists in general already inside of
the code.
Special notes for image formats
===============================
Image formats can usually skip 2) 3) and 4) from the items above. At least as long
as they do not want to build new images based on former created ones.
Special notes for new recipe types to the backend
=================================================
Note that in order to find the new recipe type on the backend when scheduling jobs,
it will most likely be necessary to teach the scheduler implementation how to match
on it. Check the expandandsort() and findfile() functions in OBS' backend:
https://github.com/openSUSE/open-build-service/tree/master/src/backend