Re: [Sdcc-user] SDCC 3.0.0 Release Candidate 1

[Raphael N. <rn...@we...>]

Hi,

> Shouldn't ./configure check for gputils?

Configure did check for gputils, but did not error out if they were not  
available due to a bug of mine (in pic16/configure.ac; the pic-port would  
never have errored out but continued to fail on build).

I attach a patch (0003-build-system.patch and 0004-generated.patch) to  
make both pic- and pic16 configure-scripts abort with error if gputils are  
not found, issuing an error like

gputils (gpasm, gplink, and gplib) are required but not found.
Either install gputils or reconfigure with --disable-pic-port and  
--disable-pic16-port.

in both ports (devices/{non-free,.}/lib/{pic,pic16}/configure).

Tested (a) without gpasm (causing abort), (b) without gpasm, with  
--disable-pic-port (aborts when configuring pic16), (c) with gpasm and  
--disable-pic-port --disable-pic16-port (succeeds without pic ports), (d)  
with gpasm and pic/pic16 ports included (succeeds as well).

After applying the patches but before commiting one might want to rerun  
"autoreconf-2.65 -fi", as the patches are generated with autoconf 2.67,  
which is reflected in aclocal.m4 .


Unrelated, build CRITICAL:
Another patch (0001-target.patch) fixes the include path generation logic  
in SDCCmain.c by avoiding use of the uninitialized 'target' -- causes  
SIGSEGV.

A last patch (0002-device-doc.patch) fixes a minor error in the list of  
supported pic14 devices.


Awaiting your approval (or commit).

Raphael