1 |
On Sun, 20 Jun 2004 17:20:11 +0100 |
2 |
Stuart Herbert <stuart@g.o> wrote: |
3 |
> In this case, it might make sense to break ht://dig into two |
4 |
> packages. Put the stuff that goes in /usr/bin et al into one |
5 |
> package, and put the stuff that goes into /var/www into a |
6 |
> separate package. |
7 |
|
8 |
Fair enough. I'll put some more thought into how this might be |
9 |
best split. |
10 |
|
11 |
> man 5 webapp.eclass should have the info you're looking for. |
12 |
|
13 |
I don't think it does, that's why I asked :) |
14 |
|
15 |
According to the man page, "Hook scripts can rely on these |
16 |
environment variables being exported: MY_HTDOCSDIR, |
17 |
MY_HOSTROOTDIR, MY_CGIBINDIR, MY_ERRORS- DIR, MY_ICONSDIR and |
18 |
MY_SERVERCONFIGDIR. They correspond to the directories listed |
19 |
in the Installation Directories section above. |
20 |
|
21 |
"Hook scripts can also rely on MY_INSTALLDIR being in the envi- |
22 |
ronment." |
23 |
|
24 |
The webapp-config script exports exactly this list. None of |
25 |
these identify the name of the virtual host being installed by |
26 |
webapp-config so, for example, the postinstall script cannot |
27 |
create a new, obviously-named search database under /var/htdig |
28 |
for that host. Only the MY_INSTALLDIR variable is close, but I |
29 |
don't think I can rely on the directory name's structure. I |
30 |
guess I could install the htdig database into a subdirectory of |
31 |
$MY_HOSTROOTDIR. Even then a configuration file needs to be |
32 |
created in a fixed location (/etc/htdig currently) and the webapp |
33 |
component needs to be modified to point to it. |
34 |
|
35 |
Perhaps the VHOST_SERVER or G_HOSTNAME variable would be useful |
36 |
to the post-install script. |
37 |
|
38 |
|
39 |
Thanks for your time, |
40 |
|
41 |
Cheers, |
42 |
Chris W |
43 |
|
44 |
-- |
45 |
gentoo-dev@g.o mailing list |