1 |
On Thu, Feb 4, 2016 at 11:27 AM, Jason Zaman <perfinion@g.o> wrote: |
2 |
> Which looks easier and nicer to you? |
3 |
> |
4 |
> NGINX_MODULES_HTTP="access auth_basic autoindex browser charset |
5 |
> fancyindex fastcgi geo gzip limit_req limit_zone map memcached proxy |
6 |
> realip referer rewrite scgi spdy split_clients ssi upstream_check |
7 |
> upstream_ip_hash userid uwsgi" |
8 |
> |
9 |
> or: |
10 |
> |
11 |
> www-servers/nginx nginx_modules_http_access nginx_modules_http_auth_basic nginx_modules_http_autoindex nginx_modules_http_browser nginx_modules_http_charset nginx_modules_http_fancyindex nginx_modules_http_fastcgi nginx_modules_http_geo nginx_modules_http_gzip nginx_modules_http_limit_req nginx_modules_http_limit_zone nginx_modules_http_map nginx_modules_http_memcached nginx_modules_http_proxy nginx_modules_http_realip nginx_modules_http_referer nginx_modules_http_rewrite nginx_modules_http_scgi nginx_modules_http_spdy nginx_modules_http_split_clients nginx_modules_http_ssi nginx_modules_http_upstream_check nginx_modules_http_upstream_ip_hash nginx_modules_http_userid nginx_modules_http_uwsgi |
12 |
|
13 |
Well, it seems like we could easily skip the _modules_http suffix, and |
14 |
the difference would not nearly be so large. |
15 |
|
16 |
The problem I have with it is that, as a user, it's yet another |
17 |
concept to grasp and configure. Instead of just working with USE flags |
18 |
(e.g. looking at use.desc or using ufed to review, configuring USE |
19 |
flags in make.conf or package.use), I now have to find/learn a bunch |
20 |
of new stuff. What packages use what USE_EXPAND things (e.g. apache |
21 |
has two), what are valid values for all of these. |
22 |
|
23 |
Where with simple USE flags, there is a single mechanism that I have |
24 |
to learn and can apply across the board; with USE_EXPAND stuff, I have |
25 |
to learn new stuff for each new package that adopts one or more |
26 |
USE_EXPAND things. |
27 |
|
28 |
So if this cosmetic expansion is the only advantage, that seems like a |
29 |
relatively limited one (e.g. it could be improved a lot just by |
30 |
formatting), and the trade-off of introducing all this extra |
31 |
complexity doesn't make that much sense to me. |
32 |
|
33 |
Cheers, |
34 |
|
35 |
Dirkjan |