team interview

[Bernd Gehrmann]

On Friday 17 May 2002 13:12, you wrote:
>
> The questions was to be understood like: does the Makefile.am look good ?
> The first version of kdevelop did insert a lot of stuff that was not needed
> for the Makefile, making it difficult to work on the Makefile.am without
> kdevelop. That made me quit kdevelop because its main interest for me was
> that it would take care of the Makefile.

Initially, KDevelop was not meant to support more than single-binary projects
for a single developer. IMO is does an excellent job in providing a simple 
user interface for such projects, and in particular using automake as backend
is an innovative approach, if you compare with other IDEs that force you into
their proprietary build system.

Of course, it has grown out of these bounds, that's why a more powerful
and less intrusive handling of Makefile.ams has been on the TODO list for 
a long time. Gideon can write *and* read Makefile.ams and modifies only
the changed lines. So even if multiple developers work on the same project
simultaneously, it is relatively easy to resolve conflicts etc.

The problem here is that it is very difficult to make all the flexibility of 
the underlying build system available through a nice user interface
(i.e. to make all-day usage simple, while making more advanced things
possible), and feedback from users will surely be helpful. Also it isn't
really helpful that KDE developers invent their own build system am_edit,
which is incompatible with vanilla automake semantics, a moving target,
and moreover totally undocumented.

Bernd.