DevGuide Comments

[David Miller]

On 12/19/2003 4:25 PM +1030, MattyT wrote:

> On Fri, 2003-12-19 at 07:59, Maxwell Kanat-Alexander wrote:
>
>> 	+ In the HTML Templates section, it says that template output should
>> start with [print "Content-type: text/html\n\n";]. I don't see that in
>> cvs-tip cgi's... Isn't it something like "print $cgi->header()"?
>
> Possibly this has changed, but the guide isn't changing until I hear
> what it should be replaced with.  It's just as likely the tip is wrong.

It should be print $cgi->header();, and you can optionally specify the
mime-type for the output as an argument to header().  $cgi is obtained from
Bugzilla->cgi; if you don't have it in scope of your current code yet
(which also means if the header is the only thing you're using $cgi for you
can just do print Bugzilla->cgi->header();)  We use the official Perl CGI
API now for everything, although we do still have backward compatibility
hacks in place to allow things like %::FORM and %::COOKIE to still work
until someone gets around to replacing them all (which will have to happen
before we can support mod_perl).  There are a few exceptions, like setting
cookies.  We have convenience functions in Bugzilla::CGI which cache your
cookies so you don't have to remember to pass them to header() for example.

>> 	+ In the "submitting your patch" section, you might want to mention
>> about what directory level to run a diff -Nru from, if it spans multiple
>> directories. It would mean that people applying patches wouldn't have to
>> worry about whether it was -p0 or -p1 or no -p at all.
>
> I'm not sure we should specify this.  Shouldn't it be obvious from the
> context?  Sometimes people might have multiple changes to their tree,
> but only want to diff one subdirectory.  Or does it work if you do it
> from Bugzilla root and specify a subdirectory?

Yes, it works if you specify a subdirectory.  Ideally, all patches should
be created from within the bugzilla directory, so that they can be applied
from that directory with a -p0 argument to patch.  A -p1 is also acceptable
if you don't have cvs handy and have to compare two bugzilla directories
(in which case you would be one level above when you make it).  If you do
it from one level up, make sure you say so on your attachment description
for the convenience of those applying the patch.  (Just "(use -p1)" on the
end is fine)

>> 	+ "first-review" and "second-review" are "review" and "super-review"
>> now, aren't they?
>
> Probably.  I'll change it when I hear the full procedure outlined.

We have "review" and "approval" at our disposal.  Review is
multiply-requestable, meaning you can request more than one.  Reviews need
to be made by people on the approved reviewers list (which is a little out
of date at the moment - I'll get that updated soon).   At least one is
required, and the person granting the first one has discretion to decide if
a second one is needed.  A second review is mandatory for security bugs.
Once a patch is reviewed and ready to check in, "approval" is requested on
the bug, which can only be granted by myself or Myk.  Once approval is
granted, the patch author can check in the patch (or someone else if the
author doesn't have cvs write access).

-- 
Dave Miller      Project Leader, Bugzilla Bug Tracking System
http://www.justdave.net/             http://www.bugzilla.org/