1 |
Tom Wijsman posted on Sat, 26 Jul 2014 00:09:58 +0200 as excerpted: |
2 |
|
3 |
> EAPI specifies what PMs need to conform to, not the other way around; |
4 |
> EAPI-0 may very well be derived from Portage, that doesn't make such |
5 |
> side features that have not been specified in EAPI-0 a part of EAPI-0. |
6 |
|
7 |
Not being around at the time, you may not know some of the history, but |
8 |
feel free to ask Ciaranm if you need a more authoritative source. |
9 |
|
10 |
The thing is, EAPI-0 was not originally completely specified, and to my |
11 |
knowledge, remains that way, because that would have been real-world |
12 |
essentially impossible to do. Instead, a convenient shortcut was taken. |
13 |
EAPI-0 was defined as what portage did at the time, with EAPI-1 for sure |
14 |
and I believe EAPI-2 at least, being defined as the the previous EAPI, |
15 |
with specifically defined changes, but with the base EAPI still only |
16 |
fuzzily defined as, basically, what portage did at the time. |
17 |
|
18 |
And since the beginning, while there have been other unapproved EAPIs not |
19 |
allowed in the main gentoo tree, because portage was and remains the |
20 |
official default PM, no EAPI has been approved for main-tree deployment |
21 |
until portage had a working implementation. |
22 |
|
23 |
So while portage can and certainly does have bugs where it doesn't meet |
24 |
EAPI requirements, particularly for behavior there since EAPI-0 and not |
25 |
specifically defined to be different in a specific EAPI since then, to |
26 |
the extent that PMS applies at all, the interpretation of PMS must still |
27 |
be read in the context of what portage did all those years ago with the |
28 |
original EAPI-0 spec, since EAPI-0 was /defined/ based on portage |
29 |
behavior at the time. |
30 |
|
31 |
Which then begs the question[1] I asked, how old /is/ this dynamic-deps |
32 |
behavior? Does it extend back to EAPI-0? My gut sense from memory as a |
33 |
user back then and now is that it does, but that's simply a gut sense I'm |
34 |
ill equipped to go back and verify. |
35 |
|
36 |
--- |
37 |
[1] Begs the question: Yes, I'm aware of the legal and philosophical |
38 |
"circular logic" usage, now generally legacy in terms of real-world use |
39 |
except in the philosophic and legal areas. I deliberately choose to use |
40 |
the phrase in the newer and now much more common sense, rhetorically |
41 |
personifying the question such that it can "beg to be asked". |
42 |
|
43 |
-- |
44 |
Duncan - List replies preferred. No HTML msgs. |
45 |
"Every nonfree program has a lord, a master -- |
46 |
and if you use the program, he is your master." Richard Stallman |