1 |
On Wed, 2003-06-11 at 08:24, D. Tuinstra wrote: |
2 |
> Michael Cummings wrote: |
3 |
> |
4 |
> > On Wed, Jun 11, 2003 at 11:37:02AM +0100, MooktaKiNG wrote: |
5 |
> >> Another good reason to have a README is becuase sometimes people |
6 |
> >> want to know more about a program. Without having to install it. |
7 |
> >> |
8 |
> >> If they can learn about the program before they install. They are |
9 |
> >> more aware of how hard/easy the installtion is. |
10 |
> > |
11 |
> > I thought that was one of the reasons the source URL shows up in a |
12 |
> > search. It's how I've always checked to see what a package does. |
13 |
> > README files, imo, are great in theory but put additional |
14 |
> > maintenance strain on a dev. Just my two euros, |
15 |
> > |
16 |
> > Mike |
17 |
> |
18 |
> As I see the original proposal, it was to give users essential |
19 |
> post-install information regarding any further steps they need to |
20 |
> take. This kind of information is different from pre-emerge info |
21 |
> that a user uses to decide the desirability of a package. |
22 |
> |
23 |
> Pre-emerge info |
24 |
> --------------- |
25 |
> Still, a small README could be valuable in some situations. For |
26 |
> example: "This package is included in portage for backwards |
27 |
> compatability for systems that have to interoperate with |
28 |
> MegaCruft's General Ledger FPS. If you don't have to do this you |
29 |
> may wish to consider <KDE-alternate> or <GNOME-alternate>. They |
30 |
> provide the same features within a modern UI, don't hog the CPU, |
31 |
> and don't require you to do manual post-emerge steps. And BTW, the |
32 |
> official site is a pathetic mass of marketing-speak, go to the |
33 |
> user-group page at <URL> instead." This shouldn't be too much of a |
34 |
> dev burden if it's optional and intended to communicate info that |
35 |
> wouldn't be found on an official web page. |
36 |
> |
37 |
> A pre-emerge README feature runs the risk of encouraging |
38 |
> editorializing (as above :-), but hey, as long as its kept |
39 |
> reasonable that's part of the fun. |
40 |
> |
41 |
> Post-emerge info |
42 |
> ---------------- |
43 |
> I'm all for it. My wish list: append each package's message to a |
44 |
> file named something like "emerge_messages_<timestamp>". These |
45 |
> would be stored in a standard location relative to the user running |
46 |
> emerge. At the end of the emerge, if the file is empty portage |
47 |
> deletes it and no mention is made of it. If the file is nonempty, |
48 |
> portage displays a standard message directing the user to read it. |
49 |
> |
50 |
> Two new emerge commands: 'einfomsg' to write a string to the message |
51 |
> file; 'einfocat' to cat a file to it. |
52 |
> |
53 |
> Messages written to the file should be no more than ~25 lines of |
54 |
> text each, with some standard header (e.g., line of dashes, ebuild |
55 |
> name, line of dashes). If 25 lines isn't enough room, the message |
56 |
> should direct the user to more instructions in |
57 |
> /usr/share/doc/<package>. |
58 |
|
59 |
It should also state which ebuild it belongs to in a very prominent |
60 |
place. I know it sounds obvious but that kind of detail is important :) |
61 |
|
62 |
-- |
63 |
Owen Ford <oford@×××.net> |