[ 
https://issues.apache.org/jira/browse/DERBY-6875?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=15170624#comment-15170624
 ] 

Rick Hillegas commented on DERBY-6875:
--------------------------------------

Thanks for picking up this issue, Bryan. I'm afraid that I can't explain the 
cgi magic because I really don't understand it myself. For the past 10 years I 
have simply been following a process which used to work.

There are two kinds of releases: mirrored releases and archived releases. 
Releases start out being mirrored and, after a year or two, are converted to 
being archived. For mirrored releases, the artifacts are automagically copied 
to lots of servers around the world. Clearly, mirroring is resource-intensive 
and the volunteer sites which host Apache releases need to garbage-collect 
stale artifacts eventually. So Apache wants us to only designate a few releases 
as mirrored. Typically, when we publish a new feature release, we then find the 
oldest mirrored release and we turn that old release into an archived release.

Both kinds of releases have html files which contain the detailed release 
notes. The two kinds of releases differ as follows:

o Mirrored releases have some extra cgi mumblings in the html file (the stuff 
that's wrong now and for which Sebb has proposed a correction). And they have a 
.cgi file which instructs the webserver to look inside the html file for those 
mumblings. The .cgi file never changes. It has exactly the same contents for 
every release.

o When the release manager changes a release from being mirrored to being 
archived, she first deletes the .cgi file. Then she edits the html file, 
replacing the cgi mumblings with hardcoded links to the archived location of 
the artifacts.

I'm not aware of any changes to this process or to the cgi magic. But I don't 
understand this cgi business, and my fingerprints are all over the release 
process by now. You can blame me.

This process goes back to the beginning of Derby. For what it's worth, I 
checked the subversion history of the 10.1.1.0 release page. Its original 
version (the version used when it was a mirrored release) has html comment 
markers around the cgi mumblings.

Your changes appear to do what Sebb wants.

+1


> Partially broken download pages
> -------------------------------
>
>                 Key: DERBY-6875
>                 URL: https://issues.apache.org/jira/browse/DERBY-6875
>             Project: Derby
>          Issue Type: Bug
>         Environment: http://db.apache.org/derby/releases/release-10.12.1.1.cgi
> http://db.apache.org/derby/releases/release-10.11.1.1.cgi
>            Reporter: Sebb
>         Attachments: site.diff, tool.diff, xdocs.diff
>
>
> The download pages have a box for choosing a mirror.
> However this is broken - it does not contain a list of mirrors.
> This is because the HTML pages are broken.
> They should contain something like the following from
> http://tomcat.apache.org/download-90.html
> {code}
> <form action="[location]" method="get" id="SelectMirror">
> <p>
>        Other mirrors: <select name="Preferred">
>          [if-any http]
>            [for http]<option value="[http]">[http]</option>[end]
>          [end]
>          [if-any ftp]
>            [for ftp]<option value="[ftp]">[ftp]</option>[end]
>          [end]
>          [if-any backup]
>            [for backup]<option value="[backup]">[backup] 
> (backup)</option>[end]
>            [end]
>          </select><input type="submit" value="Change">
> </p>
> </form>
> {code}
> However they contain:
> {code}
> <form action="[location]" method="get" id="SelectMirror">Other mirrors: 
> <select name="Preferred">
> <!--[if-any http] [for http]--><option value="[http]">[http]</option>
> <!--[end] [end]-->
> <!--[if-any ftp] [for ftp]--><option value="[ftp]">[ftp]</option>
> <!--[end] [end]-->
> <!--[if-any backup] [for backup]--><option value="[backup]">[backup] 
> (backup)</option>
> <!--[end] [end]--></select> <input type="submit" value="Change">
> </form>
> {code}
> The required templating code has been commented out, so does not work.



--
This message was sent by Atlassian JIRA
(v6.3.4#6332)

Reply via email to