Project: chromium Issues People Development process History Sign in
New issue
Advanced search Search tips
Issue 267568 Bug Deduplication Confuses hashes for words
Starred by 0 users Project Member Reported by vapier@chromium.org, Aug 2 2013 Back to list
Status: Archived
Owner: ----
Closed: Jun 26
OS: Linux
Pri: 2
Type: Bug

Blocked on:
issue 267770


Sign in to add a comment
Comment 1 by beeps@chromium.org, Aug 2 2013
Owner: beeps@chromium.org
Status: Assigned
This is because we don't follow issues and merge them into the parent yet: https://code.google.com/p/chromium/issues/detail?id=221602

If you leave the issue un duplicated we will dedupe to it but if you merge it into something else it becomes invisible to the bug filer, because there are cases like what does one do when the parent bug is closed, fixed etc. I will have this fixed shortly.

Also this will be fixed today https://code.google.com/p/chromium/issues/detail?id=267570
so we'll stop filing bugs on experimental failures.
the duplicate reports were being filed even when they were all left open.  when i got in, the majority of those bugs were not duped.  i only started duping them in the last hour or so.
Comment 3 by beeps@chromium.org, Aug 2 2013
That hash might be confusing it: Command <grep /home/user/ed5f013a79109f0d3265cceeb4b75af2b3159806 /proc/$(pgrep cryptohomed)/mounts> failed, rc=1, Command returned non-zero exit status)

as I see it deduping other failures like: https://code.google.com/p/chromium/issues/detail?id=260081&
till it was marked duplicate yesterday.

Comment 4 Deleted
Comment 5 by beeps@chromium.org, Aug 2 2013
Summary: Bug Deduplication Confuses hashes for words (was: bvt bug report generation deduplication needs work)
I initially though it was confusing < for an html escape character but it actually handles that case correctly. Whats really happening is that it transforms /home/user/ec16c7fb320f98ac27c47bab675fdd4d78e5fde4 to 
/home/user/eccfbfaccbabfdddefde because it doesn't understand where the words begin, so it errs on the side of caution and looks for the exact string in all bugs it finds, and finds nothing (it doesn't find the original hash either). when not sure it will always file a new bug.

one solution is to not put hashes that contain numbers mixed with alphabets in the exceptions/reasons. It can detect straight numbers, like addresses and certain other patterns. Another is to ignore any word like thing with numbers on white space, but that wouldn't catch /home/982473riwir/etc or something. Yet another is to teach it about common patterns like /home/usr/some-goop-we-dont-care-about, but the danger in that is deduping when we shouldn't be. 

An easy workaround would be to provide suggestions in the bug description itself, of what it might be a dupe of. A long term solution would be to provide test writers the ability to provide a regex that matches erroneous patterns that the bug filer can use to make it's decision. It's not currently capable of much intelligence (https://code.google.com/p/chromium/issues/detail?id=237131)

Renaming this bug, since the merge issue is already tracked in another one.
Comment 6 by beeps@chromium.org, Aug 2 2013
Blockedon: chromium:267770
Comment 7 by krisr@chromium.org, Sep 10 2013
Blocking: chromium:267565
Labels: bugfiling
Comment 9 by beeps@chromium.org, Feb 9 2015
Owner: ----
Status: Archived
If a tree falls in the forest and no one is around...

Archiving issues w/ no stars (external monitors), no owners, and haven't been commented on in 365 days.
Sign in to add a comment