1050255 - Offer nightly users a checkbox that indicates "this is a recent regression on Nightly"

Status: RESOLVED DUPLICATE of bug 1050259

(bugzilla.mozilla.org :: Extensions, defect)

Description

[:Gijs (he/him)]

... which should probably toggle both the "regression" keyword and a new one (nightly-regression?).

Suggested wording: "This is a new issue on Nightly that worked in the past"

We should hide this checkbox from users that aren't on Nightly (they could always add the keywords afterwards if they filed from another browser).

Comment 1

[:glob ✱]

(In reply to :Gijs Kruitbosch (Mostly gone until 22/9) from comment #0)
> We should hide this checkbox from users that aren't on Nightly (they could
> always add the keywords afterwards if they filed from another browser).

how do we determine if a user is on nightly?  i'm not aware of any way to do this anymore.

Comment 2

[:Gijs (he/him)]

(In reply to Byron Jones ‹:glob› from comment #1)
> (In reply to :Gijs Kruitbosch (Mostly gone until 22/9) from comment #0)
> > We should hide this checkbox from users that aren't on Nightly (they could
> > always add the keywords afterwards if they filed from another browser).
> 
> how do we determine if a user is on nightly?  i'm not aware of any way to do
> this anymore.

I think the only way is having some logic wrt what the latest version is, and then comparing the version string in the UA with that. In the interest of using data we have, the version options in the Firefox product seem to be in step pretty correctly, and if we assume the latest 2 versions are nightly/aurora, that seems like we should catch the 99% case here, which I think will be good enough.

(I'm happy to assume aurora users should get the same logic, considering their builds also get updated every day (after the initial uplift))

Comment 3

[Matthew N. [:MattN]]

(In reply to :Gijs Kruitbosch (Mostly gone until 22/9) from comment #2)
> (In reply to Byron Jones ‹:glob› from comment #1)
> > (In reply to :Gijs Kruitbosch (Mostly gone until 22/9) from comment #0)
> > > We should hide this checkbox from users that aren't on Nightly (they could
> > > always add the keywords afterwards if they filed from another browser).
> > 
> > how do we determine if a user is on nightly?  i'm not aware of any way to do
> > this anymore.
> 
> I think the only way is having some logic wrt what the latest version is,
> and then comparing the version string in the UA with that.

On the server side you could cache data from http://svn.mozilla.org/libs/product-details/json/firefox_versions.json to find out the Aurora version and then subtract one for Nightly. It may not be too hard to get Nightly added to that JSON file.

Comment 4

[:glob ✱]

(In reply to :Gijs Kruitbosch (Mostly gone until 22/9) from comment #2)
> I think the only way is having some logic wrt what the latest version is,
> and then comparing the version string in the UA with that. In the interest
> of using data we have, the version options in the Firefox product seem to be
> in step pretty correctly, and if we assume the latest 2 versions are
> nightly/aurora, that seems like we should catch the 99% case here, which I
> think will be good enough.

excellent thinking; thanks :)

Comment 5

[Zack Weinberg (:zwol)]

Why shouldn't there be a "This worked in a previous version of Firefox" checkbox available to everyone?  Regressions do make it all the way to release sometimes.

Comment 6

[:Gijs (he/him)]

(In reply to Zack Weinberg (:zwol) from comment #5)
> Why shouldn't there be a "This worked in a previous version of Firefox"
> checkbox available to everyone?  Regressions do make it all the way to
> release sometimes.

I'm not saying there shouldn't be, but the aim here is different. When there are serious regressions on nightly, there are often a lot of duplicates filed during the 24-48 hours that it persists (usually things don't last longer than that). It'd be useful to reduce the influx of such Fx::Untriaged bugs by categorizing "recent nightly regression" separately.

Comment 7

[Dylan Hardison [:dylan] (he/him)]

This change is going to be folded into bug 1050259