Software life cycle

jcc7 jcc7_member at pathlink.com
Fri Jul 21 08:08:52 PDT 2006


In article <e9p3b3$199m$1 at digitaldaemon.com>, MFH says...
>
>In article <e98u4v$ogc$1 at digitaldaemon.com>, Andrei Khropov says...
>>
>>Walter Bright wrote:
>>
>><skipped>
>>
>>I think Test-driven development was proposed to address this issues (or at
>>least reduce their effect).
>>
>>And D with built-in DbC and unit tests fits here quite well.
>>
>>-- 
>>
>
>unless the name of all "standard" modules is changed to 
>new.std.windows.windows.windows
>or
>std.d.stdio
>or so...
>
>Unfortunately, almost *NO* D program older than some months compiles today !
>
>(don't get me wrong, I am a *supporter* of D, but there are things that should
>be avoided in the future if D is to survive...)

This is due to improvements in DMD. Usually the the changes required are minimal
(and the compiler's error messages typically gives good hints). I don't think
that the evolution of D has been as big of a problem recently for projects that
are being maintained. 

The changes in a typical release of DMD usually provides more goodies than
gotchas.

And I think we're getting really close to a D 1.0 release which would mean that
1.0 features would be frozen and we'd only get bug fixes. That would help a lot
with the issue of code becoming obsolete.

jcc7



More information about the Digitalmars-d-bugs mailing list