• Brad Roberts (14/14) Apr 13 2007 Walter asked if I could add a way to indicate bugs that are regressions
• torhu (4/9) Apr 14 2007 Cool! Would new features that inadvertedly break backwards
• Brad Roberts (9/19) Apr 14 2007 A regression is defined as: An upgrade of the compiler resulted in code...
• Stewart Gordon (5/14) Apr 17 2007 I guess so.

Brad Roberts <braddr puremagic.com> writes:

Walter asked if I could add a way to indicate bugs that are regressions 
and have them show up in the search results in a different color.  The 
easiest way to do that with bugzilla is a new severity level.  I've 
added one called 'regression' and it will show up orange and bold in the 
search results.

For example:
     http://d.puremagic.com/issues/buglist.cgi?bug_severity=regression

I migrated all the bugs that were obviously labeled as a regression. 
There's very likely more and feel free to migrate your own bugs over as 
needed.

NOTE: this is not to say that the branching discussion thread is wrong 
or irrelevant, just a way to help prioritize issues better.

Later,
Brad

torhu <fake address.dude> writes:

Brad Roberts wrote:
 Walter asked if I could add a way to indicate bugs that are regressions 
 and have them show up in the search results in a different color.  The 
 easiest way to do that with bugzilla is a new severity level.  I've 
 added one called 'regression' and it will show up orange and bold in the 
 search results.

Cool!  Would new features that inadvertedly break backwards 
compatibility be considered regressions?  Like the ref and macro 
keywords not being disabled by -v1?

Brad Roberts <braddr puremagic.com> writes:

torhu wrote:
 Brad Roberts wrote:
 Walter asked if I could add a way to indicate bugs that are 
 regressions and have them show up in the search results in a different 
 color.  The easiest way to do that with bugzilla is a new severity 
 level.  I've added one called 'regression' and it will show up orange 
 and bold in the search results.

 
 Cool!  Would new features that inadvertedly break backwards 
 compatibility be considered regressions?  Like the ref and macro 
 keywords not being disabled by -v1?

A regression is defined as:  An upgrade of the compiler resulted in code 
that worked with a previous version of the compiler no longer does.

Now, having said that, there's a limit to where that applies.  For 
example, there's been several points in the D lifetime where something 
purposely changed and was labeled in the change log as a breaking 
change.  For the time being, I'd only consider it a regression if it 
worked for some version >= 1.000 but doesn't work in the current version.

Make sense?

"Stewart Gordon" <smjg_1998 yahoo.com> writes:

"torhu" <fake address.dude> wrote in message 
news:evqg27$17m0$1 digitalmars.com...
 Brad Roberts wrote:
 Walter asked if I could add a way to indicate bugs that are regressions 
 and have them show up in the search results in a different color.  The 
 easiest way to do that with bugzilla is a new severity level.  I've added 
 one called 'regression' and it will show up orange and bold in the search 
 results.

 Cool!  Would new features that inadvertedly break backwards compatibility 
 be considered regressions?

I guess so.

 Like the ref and macro keywords not being disabled by -v1?

It's more of a bug that they were ever enabled under -v1.

Stewart.