1 |
Hi! |
2 |
|
3 |
I've written an emake replacement that shows the progress of the make. |
4 |
I've used this for about four weeks with daily emerges, and it |
5 |
Works For Me (tm), with some (mostly) minor issues. I've attached it to |
6 |
this mail, and it looks something like this in action: |
7 |
|
8 |
[./configure output snipped] |
9 |
Guesstimating... |
10 |
Now starting make... |
11 |
2.30% |
12 |
|
13 |
Error messages are still being shown, but stdout is suppressed. |
14 |
|
15 |
Known issues |
16 |
-------------------- |
17 |
* It works by first doing a dry run of make. This adds some overhead, |
18 |
but in my experiece it is worth it. Compared to the actual make time |
19 |
it is negligible. |
20 |
* The estimate is sometimes slightly off. It is very hard to correctly |
21 |
guess the amount of work make has to do, but in most cases the |
22 |
actual make is finished within 95%-105% of the estimate, which is |
23 |
good enough for me. |
24 |
* With some ebuilds the estimate is really, really off. This goes for |
25 |
ebuilds where the ./configure does not create all necessary Makefiles |
26 |
from the beginning, but rather lets the first Makefile create new ones. |
27 |
In this case the estimate will be useless since it will only be based on |
28 |
the first Makefile. ebuilds that fall into this category are e.g. firefox and |
29 |
Xorg. ebuilds that use the ordinary autoconf/automake-tools don't have |
30 |
this problem (which is most of them). KDE works for example perfectly. |
31 |
* I've come across one ebuild that won't work with my emake replacement: |
32 |
thunderbird. So you may want to keep the original emake around :-) |
33 |
I've no clue why it won't work. Any suggestions are welcome. |
34 |
|
35 |
The script has two files: emake and progressmeter, so you need to put |
36 |
both in /usr/lib/portage/bin/ and make sure they can be executed. |
37 |
|
38 |
Any feedback is appreciated. |
39 |
|
40 |
Regards, |
41 |
Magnus Johansson |