359323 - static pages on tinderbox-stage are busted

Status: RESOLVED FIXED

(Webtools Graveyard :: Tinderbox, defect)

Description

[David Baron :dbaron: (⌚️UTC-5, no longer working on Mozilla)]

The static pages on tinderbox-stage (which we've often encouraged people to link to because they're less load on the server and they tend to load faster) are quite busted.

The following four pages should all look the same:
http://tinderbox-stage.mozilla.org/SeaMonkey/
http://tinderbox-stage.mozilla.org/showbuilds.cgi?tree=SeaMonkey
http://tinderbox.mozilla.org/SeaMonkey/
http://tinderbox.mozilla.org/showbuilds.cgi?tree=SeaMonkey

However, the first one is:
 * missing the header at the top
 * has *tons* of extra columns with seemingly everything ever associated with that tinderbox page

Sorry I didn't notice this earlier -- these pages are bookmarked / linked to a bit, but they're not easily accessible through the UI.  I think this probably should block the migration to the tinderbox-stage server.

(There are also slight differences between which columns are present on the showbuilds.cgi versions between tinderbox and tinderbox-stage, but I'd hope that's an issue that would be fixed in a final data sync -- although I'd have thought we'd be syncing that with reasonable frequency.)

Comment 1

[Jeremy Orem [:oremj]]

I'm guessing justdave would know more about this, but I'll take a look.

Comment 2

[Jeremy Orem [:oremj]]

Not sure from where this is being generated. Assigning to Dave.

Comment 3

[Dave Miller [:justdave]]

ok, this is a regression on the trunk, per discussion on #mozwebtools.  Something about $max_date not being set.

Comment 4

[cls]

Attachment: v1.0

The routines that generate the static pages expect certain globals to be set to control how much info to display.  For the cgis, these values are controlled via the form inputs.  When static pages are created from processbuild.pl, we'll just hardcode the time values.

Comment 5

[cls]

Attachment: v1.1

We now reset a few global variables (pass-by-reference anyone?) when populating the build data.  This avoids having columns from other trees bleed into the current tree.  I also cache the value of make_tree_list() since it gets called repeatedly when building static pages.

Comment 6

[David Baron :dbaron: (⌚️UTC-5, no longer working on Mozilla)]

What's currently deployed (the 09:23 copy of http://tinderbox.mozilla.org/Firefox/ ) seems to fix the extra columns problem, but not the missing header problem.

Comment 7

[David Baron :dbaron: (⌚️UTC-5, no longer working on Mozilla)]

In fact, I think it made the missing header problem worse -- the open/closed state isn't showing up anymore either.

Comment 8

[David Baron :dbaron: (⌚️UTC-5, no longer working on Mozilla)]

It's also showing a whole bunch of bogus names in the guilty column at a few points.

Comment 9

[cls]

Attachment: v1.2

This should fix the guilty column issue.  It might fix the intermittent header issue (Mozilla1.8 has it, SeaMonkey did and now doesn't).

Comment 10

[David Baron :dbaron: (⌚️UTC-5, no longer working on Mozilla)]

(I think justdave already did something to fix the guilty column issue.  Not sure if it's the same thing, though.)

Comment 11

[Dave Miller [:justdave]]

dbaron: what I did for the guilty-column was a locally-applied version of just the chunk of this patch that fixed that, based on the IRC discussion.

cls: v1.2 is live in production. open/closed state is back, but is WRONG. (/Firefox/ is showing the tree is open, but showbuilds.cgi?tree=Firefox says it's closed).

The rules still aren't showing up.

Comment 12

[Dave Miller [:justdave]]

And we have the phantom columns from other trees in there again, too. :(

Comment 13

[Dave Miller [:justdave]]

er, ignore all that, I backed out v1.1 and forgot to applied v1.2 before typing make install.  oops.  it's live now, let's see what it does.

Comment 14

[David Baron :dbaron: (⌚️UTC-5, no longer working on Mozilla)]

Seems to work pretty well now.

Hopefully this'll get checked in soon and we'll update to avoid having local patches that will confuse people later...

Comment 15

[Dave Miller [:justdave]]

Comment on attachment 244597 [details] [diff] [review]
v1.2

yup, seems good.  commit at will. :)

Comment 16

[cls]

/cvsroot/mozilla/webtools/tinderbox/processbuild.pl,v  <--  processbuild.pl
new revision: 1.61; previous revision: 1.60
done
Checking in webtools/tinderbox/showbuilds.pl;
/cvsroot/mozilla/webtools/tinderbox/showbuilds.pl,v  <--  showbuilds.pl
new revision: 1.10; previous revision: 1.9
done
Checking in webtools/tinderbox/tbglobals.pl;
/cvsroot/mozilla/webtools/tinderbox/tbglobals.pl,v  <--  tbglobals.pl
new revision: 1.46; previous revision: 1.45
done