367978 - setting up warning scraping requires manual edit of treedata.pl

Status: RESOLVED FIXED

(Webtools Graveyard :: Tinderbox, defect)

Description

[cls]

Apparently, you need to hand edit treedata.pl to do server-side warning scraping.  This is wrong for 2 obvious reasons:
1) it requires server admin shell access to enable warnings
2) treedata.pl is for tree configuration data that doesn't change

(Also, right now, the field will disappear when checksetup.pl is run because it's not in the default_treedata hash.)

We should either (a) automatically check for warnings when scraping logs for TinderPrint messages or (b) add a new section to the admin page to enable warnings for each box.

Comment 1

[cls]

Attachment: v1.0

1) This patch lets you configure boxes to be scraped for compiler warnings via the standard admin page.  

2) I rewrote the page to do a single submit for all of the per-build admin actions. 

3) I updated trim_logs() to trim scrape.dat & warnings.dat as well.

4) Switched to using |package TreeConfig| to read in scrape & warning data rather than dealing with global namespace issues in processbuild.pl.

Sidenote: warnings.pl still needs to be fixed to not rely upon /cvsroot or bonsai.

Comment 2

[Mike Taylor [:bear]]

Comment on attachment 252703 [details] [diff] [review]
v1.0

looks good - I really like the merging of tasks in admin.

Comment 3

[cls]

http://bonsai.mozilla.org/cvsquery.cgi?treeid=default&module=all&branch=HEAD&branchtype=match&dir=mozilla%2Fwebtools%2Ftinderbox&file=&filetype=match&who=cls%25seawood.org&whotype=match&sortby=Date&hours=2&date=explicit&mindate=2007-01-25+12%3A12&maxdate=2007-01-25+12%3A14&cvsroot=%2Fcvsroot