From: Alan W. I. <ir...@be...> - 2003-09-03 21:01:27
|
This came up in private discussions with Rafael, and we thought we should get the list involved in the discussion. Alan: > "NOBRAINDEAD appears in a number of places in the source tree, but it is essentially totally unsupported in configure.ac and sysloc.in." Rafael: > "The Right Way (TM) of doing this is by adding an AC_DEFINE(NOBRAINDEAD,...) to configure.ac and having its value being controlled by a configure option like --enable-nobraindead. The current situation is not satisfactory, since plConfig.h.in (which is CVS controlled) must be edited to change this #define. We could issue a *_big_* warning if the user specify that switch." My first tendency is to say just leave it in its current (unsupported-by-autotools) state since it is such an exotic option that is rarely used (never by me, for example). However, I agree with Rafael's implicit point, that having half-supported options around like this can be nothing but a pain for future maintenance/understanding of what is going on with NOBRAINDEAD. I am willing to make the changes (essentially following what is done for the with_freetype logic) if the rest of the list feel this is worthwhile. Or would it be better to remove NOBRAINDEAD completely from the source tree. Has anybody actually used it in the last 5 years? Alan |