1 |
On Wed, Jan 10, 2001 at 01:57:38AM -0700, Bill Anderson wrote: |
2 |
|
3 |
> > That's the nifty thing about the portage system--it takes care of |
4 |
> > creating the packing list. |
5 |
|
6 |
> say, that is nifty. |
7 |
|
8 |
We think so too :) |
9 |
|
10 |
> Hmm, excuse my newness to gentoo (not installed until I get the drive |
11 |
> backed up tomorrow or so (I hope)) ... but does this mean that hyphens |
12 |
> are not valid in an ebuild name? |
13 |
|
14 |
Hyphens are valid, but ebuild has a special "version syntax" so that it can |
15 |
parse the version numbers of a package and compare them. Here are the basic |
16 |
rules: |
17 |
|
18 |
package names can have any number of hyphen-separated words, like so: |
19 |
|
20 |
foo-bar-oni-ville (this is a valid package name) |
21 |
|
22 |
then, to tag the version on it, you add another hyphen and a version string. |
23 |
The version strings have a special format: |
24 |
|
25 |
4.0.0 valid |
26 |
4.0.1 valid |
27 |
4.01.1109235.12350923.329832932.3232983 valid |
28 |
4.01_alpha valid |
29 |
4.02_alpha17 valid |
30 |
7.1.3_beta valid |
31 |
7.1.3.5_beta5 valid |
32 |
(we support the _pre tag in the same way) |
33 |
4.0a is valid |
34 |
|
35 |
7.1.3-pre8 is *not* valid (should be 7.1.3_pre8) |
36 |
4.0aa is *not* valid (we support revisions "a" through "z" -- a single character) |
37 |
2.4.0-test10 is *not* valid (we don't support "test" just yet, and if we did the correct |
38 |
format would be 2.4.0_test10) |
39 |
|
40 |
OK, so that's the basic intro to the version string. Then, there's an optional |
41 |
package "rev" that defaults to zero if you leave it off. However, if you want to |
42 |
specify a "rev", you add a -r<revnum>. So, a final package/version/rev name could |
43 |
be something like: |
44 |
|
45 |
foo-bar-oni-ville-4.0.3.1.34.4.3.3_alpha76-r12 |
46 |
|
47 |
Amazingly, Portage can parse this and determine if it's newer or older than |
48 |
foo-bar-oni-ville-4.012323420.9393.3.3.3.3.3.232.5.3_alpha73-r11 |
49 |
|
50 |
However, we normally don't get that complicated. |
51 |
|
52 |
> As compared to a prerelease test kernel? |
53 |
> As someone who has been through many release cylces, I would have to |
54 |
> dosagree, if that is your position. |
55 |
> |
56 |
> Justin |
57 |
> As far as 'ac' those are Alan cox patched kernel. IME, they've been much |
58 |
> better than the standard kernel. Alan Cox is basically the second in |
59 |
> command in Linuxville, eventually the ac patches become merged (usually, |
60 |
> but not always). |
61 |
|
62 |
The most important thing to us is *ReiserFS filesystem stability*. We wait |
63 |
until we can ensure that ReiserFS works well and doesn't dice up everyone's |
64 |
files. Right now, Achim is testing 2.4.0-ac3 with patches and making sure |
65 |
it's stable. |
66 |
|
67 |
-- |
68 |
Daniel Robbins <drobbins@g.o> |
69 |
President/CEO http://www.gentoo.org |
70 |
Gentoo Technologies, Inc. |