Hello Carsten,
I've looked into this, and some things I can do in Confluence, other
things I cannot. :)
I can generate an HTML page with the variables you described, and I
think it is convenient to have such a page in Confluence so we can
easily edit it for every release we do (with all the subprojects we
will be doing quite a bit of releasing).
I cannot add a cgi script in the same folder, neither can I set the
script bit if I add such a script to a page as an attachment (in
which case it will be created in the PAGENAME.data folder, so that's
one level below the actual page anyway).
I currently see a couple of options:
a) Use a download.cgi script that is modified a bit so the html file
it parses does not need to be in the same directory (but in one we
can supply in the script). Unfortunately, my python/cgi skills are
not sufficient to see if this is easily possible. I would like:
http://host/download.cgi?template=http://otherhost/downloads.html
b) Deploy the download.cgi and download.html files somewhere else,
and include them in a Confluence page with (for example) an iframe.
This would at least give us some freedom in modifying the site
through Confluence.
Option a) would be the best (at least for now) if that can be done
easily.
By the way, is there some consensus about how to make older versions
available? I mean, do we need a "download" page per release? Ideally
I think we would like to automate this as much as possible. Does it
make sense to generate the download.cgi and download.html as part of
a "release" target, together with all other artifacts, so we can just
copy all of them to dist/ (or dist/versionX.Y.Z/) and go from there?
Expanding on that "automating" point, it would be nice if the
artifacts we generate using our continuous build could be "released"
easily. This would go for both our source and binary releases, but
probably also for "generated" documentation related to those releases
(javadoc, unit test and coverage results, etc.).
Another question, when I log in to people.apache.org, what folder
should I go to to modify files that appear in www.apache.org/dist/
felix/ ? I thought I found the directory (/www/www.apache.org/dist/
felix/) but stuff I put in there did not show up (unless it takes
some time for them to show up?).
Greetings, Marcel
On Jul 31, 2007, at 14:11 , Carsten Ziegeler wrote:
My first idea was to replace to confluence page completly with a
static
html page...but if we get the cgi working inside confluence that would
be much easier.
The current solution uses (as installed in other projects) uses two
files:
* download.cgi
* download.html
The first one is the cgi which is called, so the "download" link
points
to download.cgi. The cgi is pretty simple:
!/bin/sh
# Wrapper script around mirrors.cgi script
# (we must change to that directory in order for python to pick up the
# python includes correctly)
cd /www/www.apache.org/dyn/mirrors
/www/www.apache.org/dyn/mirrors/mirrors.cgi $*
This cgi in turn includes the download.html (in fact it includes an
html
page of the same name as the cgi - so if we would call the cgi
dw-felix.cgi, the html has to be named dw-felix.html).
Inside the html page, you can use some variables like [preferred]
to get
the actual selected mirror, so a download url is then
[preferred]/felix/something....
We also need a form to change the mirror. (I'll not paste the code for
now, but I can give it to you if you want).
Do you think that this is possible with using confluence?
