Fwd: REST APIs, and Tags

Gervase Markham gerv at mozilla.org
Fri Feb 12 12:23:51 UTC 2010

On 11/02/10 13:32, Frédéric Buclin wrote:
> My two main worries are 1) that it's harder to remember the exact syntax
> of each label/tag. With our current design, you have each field
> separately, with a list of values you can choose from. That's not prone
> to errors.

I think it is, not just the same sort of "errors".

The difficulty is, those of us with highly organized minds have a
tendency to think that as long as people are provided with the ability
to organize everything perfectly, then things will end up organized
perfectly. Sadly, that isn't the case.

The sort of "errors" you get with the current system are people not
setting a field, or setting it to the wrong value, or setting all the
fields to the right values but the person who is supposed to be finding
the bug to fix it looking for different values instead.

> Also, 2) with our current design, I know exactly where to look at for a
> given information, because a field is always at the exact same place in
> a bug report. With tags, I fear my eyes have to scan the whole list of
> tags all the time to know if such or such attribute is present. 

That's a UI question. A long undifferentiated list is not the only
possible UI for a tag-based system.

But also, your eye is trained to skip past all the fields that e.g. the
Bugzilla project never or rarely uses, such as Platform and OS. However,
these fields increase visual complexity and unapproachability.
Additionally, their global nature means that most of the values are
irrelevant for most of the software tracked in b.m.o.

If Platform and OS information were only present when it was relevant,
that would draw attention to the fact that the bug was platform or

> Gerv, I'm not saying Google is wrong and Bugzilla is the right way to
> display or use metadata. I'm saying that you cannot use labels
> everywhere and all the time. 

I'm not arguing for "all the time". What I'd really like to see, to
being with, is a first-class tagging implementation. Then people could
decide what to use them for. :-)

> So it's rather
> a UI thing than a real rethink of the whole way Bugzilla works.

Again, I'm not arguing for us to rewrite Bugzilla from the ground up as
the Google bug tracker :-) What I want is for us to implement a proper
tagging system - which takes what we've learned from the status
whiteboard, keywords, flags and custom fields, rolls in what the rest of
the web has learned from their implementations of tagging, applies it to
the Bugzilla context (so e.g. public and private tags) and implements it
in a way which is fast (so a tags table and an integer-based search

dev-apps-bugzilla mailing list
dev-apps-bugzilla at lists.mozilla.org

More information about the developers mailing list