Will 1.45 incorporate active regression tickets?
Will 1.45 incorporate active regression tickets? I.e., report {30} <https://svn.boost.org/trac/boost/report/30> Several of these could be incorporated without too much work. I think a scan of these would be worth the release managers' time.
On Oct 31, 2010, at 6:32 PM, Jim Bell wrote:
Will 1.45 incorporate active regression tickets? I.e., report {30} <https://svn.boost.org/trac/boost/report/30>
Several of these could be incorporated without too much work. I think a scan of these would be worth the release managers' time.
While I'm not a release manager, I looked at a few of them; and I'm wondering if there's a disagreement here about what constitutes a "regression". To me, that means that the code has "regressed", or gotten worse. Something that worked in a previous release no longer works - that's a "regression" Something that didn't work before is a "bug". If a new test is added, and it fails on some platforms - that's a "bug". Something that's being tested on a new OS/CPU/compiler and doesn't work - that's a "bug". It's possible, however, that other people are using these terms to mean different things. Anyone else want to chime in? -- Marshall
On 10/31/2010 08:11 PM, Marshall Clow wrote:
... elision by patrick >8 ... While I'm not a release manager, I looked at a few of them; and I'm wondering if there's a disagreement here about what constitutes a "regression".
To me, that means that the code has "regressed", or gotten worse. Something that worked in a previous release no longer works - that's a "regression" You're right. It has two common meanings in bugs:
1) a trend or shift toward a lower or less perfect state i.e. there's a new bug that didn't use to be there. 2) a movement backward to a previous and especially worse or more primitive state i.e. a bug that was previously fixed is back, or a new feature has disappeared. The second one can sometimes be deceptive when a new bug appears with the same symptoms as an older bug. People are quick to assume a regression, but it could be a new bug. All bugs are bugs whether they're regressions or not. Patrick
On 1:59 PM, Patrick Horgan wrote:
On 10/31/2010 08:11 PM, Marshall Clow wrote:
... elision by patrick >8 ... While I'm not a release manager, I looked at a few of them; and I'm wondering if there's a disagreement here about what constitutes a "regression".
To me, that means that the code has "regressed", or gotten worse. Something that worked in a previous release no longer works - that's a "regression" You're right. It has two common meanings in bugs:
1) a trend or shift toward a lower or less perfect state i.e. there's a new bug that didn't use to be there. 2) a movement backward to a previous and especially worse or more primitive state i.e. a bug that was previously fixed is back, or a new feature has disappeared.
[...] All bugs are bugs whether they're regressions or not.
I was defining it more tangibly as yellow on the regression matrix, but, yes, that could be from something going from green to yellow (your definition) or something going from not-being-there-before to yellow (i.e., a new platform). Could someone nail down the definition in the appropriate place (Wiki)? This would be important for the new wave of volunteers being discussed.
On Nov 1, 2010, at 6:04 AM, Jim Bell wrote:
On 1:59 PM, Patrick Horgan wrote:
On 10/31/2010 08:11 PM, Marshall Clow wrote:
... elision by patrick >8 ... While I'm not a release manager, I looked at a few of them; and I'm wondering if there's a disagreement here about what constitutes a "regression".
To me, that means that the code has "regressed", or gotten worse. Something that worked in a previous release no longer works - that's a "regression" You're right. It has two common meanings in bugs:
1) a trend or shift toward a lower or less perfect state i.e. there's a new bug that didn't use to be there. 2) a movement backward to a previous and especially worse or more primitive state i.e. a bug that was previously fixed is back, or a new feature has disappeared.
[...] All bugs are bugs whether they're regressions or not.
I was defining it more tangibly as yellow on the regression matrix, but, yes, that could be from something going from green to yellow (your definition) or something going from not-being-there-before to yellow (i.e., a new platform).
Could someone nail down the definition in the appropriate place (Wiki)? This would be important for the new wave of volunteers being discussed.
I would be happy to - once I'm sure that we're all in agreement. -- Marshall
On 1:59 PM, Marshall Clow wrote:
On Nov 1, 2010, at 6:04 AM, Jim Bell wrote:
On 1:59 PM, Patrick Horgan wrote:
On 10/31/2010 08:11 PM, Marshall Clow wrote:
... elision by patrick >8 ... While I'm not a release manager, I looked at a few of them; and I'm wondering if there's a disagreement here about what constitutes a "regression".
To me, that means that the code has "regressed", or gotten worse. Something that worked in a previous release no longer works - that's a "regression" You're right. It has two common meanings in bugs:
1) a trend or shift toward a lower or less perfect state[...]
[...] All bugs are bugs whether they're regressions or not. I was defining it more tangibly as yellow on the regression matrix, but, yes, that could be from something going from green to yellow (your definition) or something going from not-being-there-before to yellow (i.e., a new platform).
Could someone nail down the definition in the appropriate place (Wiki)? This would be important for the new wave of volunteers being discussed. I would be happy to - once I'm sure that we're all in agreement.
I don't insist that anything on the regression matrix must be called 'regression', but does an unexplained failure on the regression matrix deserve special attention? I think so, as unfair as that is when new platforms are added. It's something we *know should* work.
On 11/01/2010 10:45 AM, Jim Bell wrote:
... elision by patrick >8 ... I don't insist that anything on the regression matrix must be called 'regression', but does an unexplained failure on the regression matrix deserve special attention? I think so, as unfair as that is when new platforms are added. It's something we *know should* work. That's a great definition of a regression, when you know that something should work, because it already worked, but somebody did something to it and now it doesn't.
(Not to include the other case where I know the code I wrote SHOULD work, but it doesn't and never has, because it does what I said instead of reading my mind.) Patrick
On 1:59 PM, Patrick Horgan wrote:
On 11/01/2010 10:45 AM, Jim Bell wrote:
... elision by patrick >8 ... I don't insist that anything on the regression matrix must be called 'regression', but does an unexplained failure on the regression matrix deserve special attention? I think so, as unfair as that is when new platforms are added. It's something we *know should* work. That's a great definition of a regression, when you know that something should work, because it already worked, but somebody did something to it and now it doesn't.
(Not to include the other case where I know the code I wrote SHOULD work, but it doesn't and never has, because it does what I said instead of reading my mind.)
Patrick
Agreed.
On Nov 1, 2010, at 6:04 AM, Jim Bell wrote:
On 1:59 PM, Patrick Horgan wrote:
On 10/31/2010 08:11 PM, Marshall Clow wrote:
... elision by patrick >8 ... While I'm not a release manager, I looked at a few of them; and I'm wondering if there's a disagreement here about what constitutes a "regression".
To me, that means that the code has "regressed", or gotten worse. Something that worked in a previous release no longer works - that's a "regression" You're right. It has two common meanings in bugs:
1) a trend or shift toward a lower or less perfect state i.e. there's a new bug that didn't use to be there. 2) a movement backward to a previous and especially worse or more primitive state i.e. a bug that was previously fixed is back, or a new feature has disappeared.
[...] All bugs are bugs whether they're regressions or not.
I was defining it more tangibly as yellow on the regression matrix, but, yes, that could be from something going from green to yellow (your definition) or something going from not-being-there-before to yellow (i.e., a new platform).
Could someone nail down the definition in the appropriate place (Wiki)? This would be important for the new wave of volunteers being discussed.
I've updated the <https://svn.boost.org/trac/boost/wiki/TicketWorkflow> page on the wiki with a first cut. Feedback solicited/welcomed -- Marshall
On Nov 1, 2010, at 6:04 AM, Jim Bell wrote: ... elision by patrick ... I've updated the<https://svn.boost.org/trac/boost/wiki/TicketWorkflow> page on the wiki with a first cut. Feedback solicited/welcomed I think it's clear.
Patrick
participants (3)
-
Jim Bell -
Marshall Clow -
Patrick Horgan