pkgsrc-Users archive

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index][Old Index]

Re: wm/ctwm without cmake

> Is there any point to this thread now? I'm not getting paid (enough)
> for pkgsrc work to shoulder in-pkgsrc build systems
> which will quickly diverge from upstream, making the work more
> difficult in the long run than necessary. Similar to what Jörg wrote
> before.
> Do you just want to rant about CMake, or what's the point of this now?

First, let me say that trying to enforce a policy to bend
third-party packages away from cmake "against their will" is most
probably a waste of time, so I'll side with Joerg on this one.

That being said, as a cmake-newbie I have done battle with
cmake-using packages in an attempt to find out "where is the
error", and it is very often far from trivial:

 * The dependencies and conditionals which influence the setting
   of a given variable is often Very Difficult to trace, perhaps
   primarily because we don't have any tooling which aids with
   this process.  I think I saw a pointer to a "cmake debugging
   tool" somewhere, but I don't think we have that available.
   Windows only?  (It looked kind of curses-like, so I would have
   thought it should in principle not be impossible on our

 * The cmake code fragments which influence the setting of a
   given variable can come from "anywhere".  This includes "from
   already installed packages", either from the cmake package
   itself, but can also come from other packages.  See the point

 * I get a sense that there is an adopted style or explicit
   desire to use maximum levels of indirection when defining or
   setting a variable.  Therefore, manually tracing all the
   myriad of different conditionals and dependencies which
   influence a setting can quickly lead you to get lost in the
   weeds.  See the first point above.

So ... I'm not a fan of debugging cmake-using packages,
especially when something goes wrong.


- Håvard

Home | Main Index | Thread Index | Old Index