1 |
Quoting Stuart Herbert <stuart@g.o>: |
2 |
|
3 |
> First off - thank you Troy for getting this moving. Nice one ;-) |
4 |
|
5 |
+1 :-) |
6 |
|
7 |
> I think this should just be /usr/share/webapps/<application>/ |
8 |
> |
9 |
> Sorry, but I very strongly disagree. Not every web application needs all |
10 |
> its files to be accessible via a web browser. |
11 |
|
12 |
Granted, but this means more changing in the core or config files of the |
13 |
particular application. |
14 |
|
15 |
> /usr/share/webapps/htdocs/ for files that should be accessible via a web |
16 |
> browser. This leaves room for any other files (PHP scripts are the one |
17 |
> that comes to mind for me) to go into other sub-directories if required. |
18 |
> |
19 |
> (Confession: I do have a personal interest. I write web apps |
20 |
> professionally, and none of them have the PHP files available for |
21 |
> serving out via the web browser.) |
22 |
|
23 |
Same here. |
24 |
|
25 |
> I'd prefer /etc/webapps/<application>/, again for future flexibility |
26 |
> where a web application has more than one configuration file. I don't |
27 |
> mind frigging apps to look in more than one place for a config file. I |
28 |
> do mind patching apps to use just the one file. |
29 |
|
30 |
You want to mix the apache config block with other configuration files that |
31 |
come with the application? Seems messy to me. Especially if we're going |
32 |
to be supporting multiple webservers (creating server-specific config |
33 |
files). Perhaps another directory where you can separate the |
34 |
application-specific config files from server-specific ones. Ideas? |
35 |
|
36 |
> The less we have to patch an app, the less work we're going to make for |
37 |
> ourselves - especially on the support side. |
38 |
|
39 |
Yes! Which is why a patch is probably not always appropriate -- sed is |
40 |
more resilient to pieces of configuration moving around upstream. |
41 |
|
42 |
> If Robin doesn't beat me to it, I'll write a script that we can add to |
43 |
> Portage to standardise getting this information. |
44 |
|
45 |
What did you have in mind to achieve this? |
46 |
|
47 |
> W.r.t. PHP detection, we have two other problems: |
48 |
> |
49 |
> a) Depending on virtual/php is useless, because it doesn't guarantee that |
50 |
> mod_php is installed (could be that PHP/CLI is installed) - unless you've |
51 |
> already fixed this Robin? Anyway, it doesn't tell you what version of |
52 |
> PHP is installed - and some apps need specific versions of PHP (yes, |
53 |
|
54 |
I'll leave this to you all as you seem to have hashed this out already |
55 |
between you. |
56 |
|
57 |
> What shape are other languages in for their dependencies? Not every |
58 |
> webapp is going to be written in PHP ... |
59 |
|
60 |
Many apps are CGIs. Others can of course be mod_perl, mod_python, java, |
61 |
you name it. I see no problems using the eclass with these because it was |
62 |
written with flexibility in mind. Like I've said in the past. My initial |
63 |
idea for this was to fix nut and apcuspd to cleanly install their CGI |
64 |
components. |
65 |
|
66 |
> Well, write access to directories under htdocs at any rate. TikiWiki - |
67 |
> which we're looking to use for a Gentoo website - has this annoying |
68 |
> feature. |
69 |
|
70 |
What bothers me though, is that having any directory under /usr writeable |
71 |
is really bad form (I made a policy at work for all servers -- configure a |
72 |
system where /usr can be mounted read-only). If at all possible, I'd like |
73 |
to have these directories created under something like /var/lib/ and |
74 |
symlinked back to where the app needs to write. If not symlinked then an |
75 |
Apache alias (or whatever is equivalent to other servers). |
76 |
|
77 |
> You use sed, I'd apply patches, and I'm sure there's other ways too that |
78 |
> people would like to use. So long as they get the right result, I don't |
79 |
> think we need to standardise on the approach. |
80 |
|
81 |
See above. But I agree, we _can't_ standardize on the approach simply |
82 |
because it depends on what needs to be done. I imagine that there will be |
83 |
some apps where we would just have complete config files living in |
84 |
${FILESDIR} that get installed over the ones that come with the package. |
85 |
Having said that, I say we _should_ standardize on installation of packages |
86 |
from the same family (i.e. Horde). |
87 |
|
88 |
>> Standardization should always be applauded. :-) |
89 |
> |
90 |
> Urgh ;-) If everything in life was standardised, we'd be running RedHat, |
91 |
> not busy upsetting the apple-cart with this upstart project ;-) |
92 |
|
93 |
Having run redhat for the past 4 years, I can safely say that that pile of |
94 |
stuff strewn loosely together with twine and masking tape into a |
95 |
nightmarish packaging system is far from standardized. |
96 |
|
97 |
> First off, where are webapp ebuilds going to live in Portage? I'd like |
98 |
> to see a set of 'web-???' categories myself, so that web apps are |
99 |
> clearly visible and easy to find. I don't think they should disappear |
100 |
> under 'net-www' or into various 'app-???' directories. |
101 |
|
102 |
Forgot to bring this up. Can't agree more. Web-* makes sense to me. |
103 |
There's a slew of things that can go into web-mail and web-net just to |
104 |
start. |
105 |
|
106 |
> Secondly, are we going to establish a webapps herd to look after the |
107 |
> packages that will be added? If so, feel free to add me to the list of |
108 |
> maintainers. If not, then what solution do you suggest? |
109 |
|
110 |
A herd makes sense. I'll leave this to the higher-ups thought. :-) |
111 |
|
112 |
> I think the ebuilds for web apps should be ENTIRELY webserver neutral. |
113 |
> This is why I've done nothing with Max's eclass myself. There should be |
114 |
> userland tools for creating an instance of an app under Apache or |
115 |
> whatever. Yes, I know that most people will run Apache, and not care |
116 |
> about this, but Gentoo's supposed to be about configurability. We have |
117 |
> a chance to get some solid foundations in - I say let's get digging. |
118 |
|
119 |
As I've said before many times, I'd like to see the eclass grow into |
120 |
something that can be used with all the webservers we have in portage. I |
121 |
don't know if userland tools can be flexible enough to create config blocks |
122 |
for all the apps that we're going to have. An ebuild knows enough of the |
123 |
application to pass the necessary information to the eclass to create |
124 |
whatever config is needed. At the same time, we have to be careful to not |
125 |
balloon this into something unmaintainable. Which is why it may be best to |
126 |
start this as apache-only (as it is the more popular of the webservers). |
127 |
Get everything converted over and working and only then add support for |
128 |
others. |
129 |
|
130 |
--mk |
131 |
|
132 |
-- |
133 |
gentoo-dev@g.o mailing list |