1 |
On Tue, 26 Oct 2010 18:13:25 +0200 |
2 |
Fabian Groffen <grobian@g.o> wrote: |
3 |
|
4 |
> On 26-10-2010 17:39:04 +0200, Tomáš Chvátal wrote: |
5 |
> > So since I would like to have this script replace current eshowkw |
6 |
> > one I want to ask you for kind help on that script and review it |
7 |
> > and suggest improvements. |
8 |
|
9 |
Great work! |
10 |
|
11 |
> I think you should implement some means of selecting which archs you |
12 |
> want to plot. Preferably through some config file thing. I'd think |
13 |
> of specifying per arch, and or using predefined sets, e.g. to remove |
14 |
> all prefix arches, since they spoil any output with huge unreadable |
15 |
> matrices. |
16 |
|
17 |
+1. |
18 |
|
19 |
In fact my adapted version of the /old/ eshowkw already filters |
20 |
everything after the line '^# Prefix keywords$' in profiles/arch.list, |
21 |
simply because the entire bunch arch.list entries makes the output |
22 |
broader than what happens to be my default $COLUMNS of 87, which breaks |
23 |
the layout, and because I hardly ever need to know about the prefix |
24 |
keywording. |
25 |
|
26 |
Alternatively to arch.list, the script might be capable of parsing |
27 |
profiles.desc in order to establish which arches can go stable, by |
28 |
using information from the first and third columns, as well as getting |
29 |
the best option out of the third column, so 'stable' for an arch which |
30 |
also has 'dev' or 'exp' profiles. That could be used below. |
31 |
|
32 |
> Have you ever played with rotating the view? Some keywords are |
33 |
> awfully long, like sparc64-solaris. Most versions are shorter. |
34 |
> In case the number of versions is small, you could even have the |
35 |
> versions in columnar style (e.g. not vertical, just a column spanning |
36 |
> a couple of chars), such that it is much better readable. |
37 |
|
38 |
+1. |
39 |
|
40 |
> Maybe consider a raw, tab, csv alike output format, for easy parsing |
41 |
> by other scripts? Could perhaps handy to render it alternatively as |
42 |
> xml, latex, etc. |
43 |
|
44 |
+1. |
45 |
|
46 |
Combined with the arch masking option, this could easily show a package |
47 |
maintainer which arches to call on a stabilisation request. All too |
48 |
often an arch is forgotten or needlessly added, when a script parsing |
49 |
arch.list/profiles.desc could easily sort it out and even print a list |
50 |
of arch@g.o aliases. |
51 |
|
52 |
|
53 |
jer |