| 1 |
Hi all, |
| 2 |
|
| 3 |
ebuild.sh is a mess. There are a lot of functions scattered to the four |
| 4 |
winds. Searching for a function in ebuild.sh takes a lot of time, and it |
| 5 |
is very tiring to scroll down huge chunks of totally unrelated functions |
| 6 |
just to find the next one which one is after. The current layout also |
| 7 |
makes it very hard to find the spots where new functions/variables are |
| 8 |
loaded from other files, what gets overwritten when profiles/eclasses |
| 9 |
are sourced etc. |
| 10 |
|
| 11 |
I would like to change that, so I tried to categorize all the functions |
| 12 |
we have in ebuild.sh. I ended up with this setup: |
| 13 |
|
| 14 |
Sandbox: |
| 15 |
addread() |
| 16 |
addwrite() |
| 17 |
adddeny() |
| 18 |
addpredict() |
| 19 |
lchown() |
| 20 |
lchgrp() |
| 21 |
|
| 22 |
Unknown?: |
| 23 |
esyslog() |
| 24 |
|
| 25 |
Debugging: |
| 26 |
register_die_hook() |
| 27 |
diefunc() |
| 28 |
dump_trace() |
| 29 |
debug-print() |
| 30 |
debug-print-function() |
| 31 |
debug-print-section() |
| 32 |
|
| 33 |
Internal helpers: |
| 34 |
strip_duplicate_slashes() |
| 35 |
remove_path_entry() |
| 36 |
|
| 37 |
Action functions: |
| 38 |
dyn_setup() |
| 39 |
dyn_unpack() |
| 40 |
dyn_clean() |
| 41 |
dyn_compile() |
| 42 |
dyn_test() |
| 43 |
dyn_install() |
| 44 |
dyn_preinst() |
| 45 |
dyn_help() |
| 46 |
|
| 47 |
Abort handlers: |
| 48 |
abort_handler() |
| 49 |
abort_compile() |
| 50 |
abort_unpack() |
| 51 |
abort_test() |
| 52 |
abort_install() |
| 53 |
killparent() |
| 54 |
|
| 55 |
Default functions for ebuilds: |
| 56 |
pkg_setup() |
| 57 |
pkg_nofetch() |
| 58 |
src_unpack() |
| 59 |
src_compile() |
| 60 |
src_test() |
| 61 |
src_install() |
| 62 |
pkg_preinst() |
| 63 |
pkg_postinst() |
| 64 |
pkg_prerm() |
| 65 |
pkg_postrm() |
| 66 |
pkg_config() |
| 67 |
|
| 68 |
Ebuild helpers: |
| 69 |
into() |
| 70 |
insinto() |
| 71 |
exeinto() |
| 72 |
docinto() |
| 73 |
insopts() |
| 74 |
diropts() |
| 75 |
exeopts() |
| 76 |
libopts() |
| 77 |
keepdir() |
| 78 |
unpack() |
| 79 |
econf() |
| 80 |
einstall() |
| 81 |
use() |
| 82 |
usev() |
| 83 |
useq() |
| 84 |
has_version() |
| 85 |
portageq() |
| 86 |
best_version() |
| 87 |
use_with() |
| 88 |
use_enable() |
| 89 |
gen_wrapper() |
| 90 |
check_KV() |
| 91 |
inherit() |
| 92 |
EXPORT_FUNCTIONS() |
| 93 |
|
| 94 |
Internal eclass functions: |
| 95 |
newdepend() |
| 96 |
newrdepend() |
| 97 |
newpdepend() |
| 98 |
do_newdepend() |
| 99 |
|
| 100 |
I moved them into own files within a subdir in bin/, called modules, and |
| 101 |
source these files at the right time, so that nothing changes |
| 102 |
functionality wise. The global scope code I left untouched, except for a |
| 103 |
bunch of exports that set sane defaults for the do*/new* helpers. |
| 104 |
|
| 105 |
This cuts down ebuild.sh to 403 lines, which makes it far more readable, |
| 106 |
IMO. |
| 107 |
|
| 108 |
Doing the refactoring I also saw that there is a esyslog() in |
| 109 |
isolated-functions.sh and one in ebuild.sh. The latter, which does |
| 110 |
nothing but return zero, overwrites the former one. Is this intended or not? |
| 111 |
|
| 112 |
I also set all the functions that aren't meant to be overridden by saved |
| 113 |
env, profile bashrc, portage bashrc or ebuilds/eclasses readonly. If an |
| 114 |
ebuild would ever try to re-define a function which is internally used |
| 115 |
by portage, it would fail sourcing, which is a good thing, IMO. |
| 116 |
|
| 117 |
So, if noone objects, I would like to push the attached patch into SVN. |
| 118 |
|
| 119 |
-- |
| 120 |
Kind regards, |
| 121 |
|
| 122 |
Simon Stelling |
| 123 |
Gentoo/AMD64 developer |
Archives