* support/regression/Makefile.in: fixed distclean for cases dir
I just found out that the bug in boost was fixed in version 1.79.
In the past, when a problem with binutils came up, I usually upgraded to the latest release of GNU binutils. To do so, I first looked at the diff from the current version (i.e. compared current sdbinutils to the version of GNU binutils it is based on), then made corresponding changes to the current GNU binutils (remove targets irrelevant for SDCC, add the little bit of extra stuff SDCC needs).
https://sourceforge.net/p/sdcc/feature-requests/940/ [feature-requests:#940]
I would not make it an error. Nothing prevents some mfg to create an mcs-51 compatible chip with more or less sfr space. Further __data __at() is also useful for some sfr that does not need to be treated as volatile. And more in general: don't limit the user unless you have to. You cannot let the compiler ignore an error. Using __at() is an advanced tool and comes with responsibilities. If you cannot handle that you should not use it. You should always ask yourself: do I really need __at() here?...
https://sourceforge.net/p/sdcc/feature-requests/940/ [features:#940]
Far pointers and maybe a huge memory model for stm8
Suboptimal but otherwise correct code is never a bug.