[RPL/2] compile error

[BERTRAND Joel]

DEMAINE Benoit-Pierre a écrit :
> On 19/05/10 17:43, BERTRAND Joel wrote:
>> DEMAINE Benoit-Pierre a écrit :
>>> doublehp-dhp/rpl2-4.0.0_pre6-r2 used to work fine (pre6 for pre-release,
>>> -r2 is my internal ebuild version, does not affect downloaded version),
>>> and could be used as CLI.
>>>
>>> doublehp-dhp/rpl2-4.0.14 produce the following message:
>>> (I have to use quotes to ensure alias over-riding)
>>>
>>> dhp at uranus ~ $ "rpl" -i
>>> +++RPL/2 (R) version 4.0.14 (Wednesday 05/19/10, 14:23:18 CEST)
>>> +++Copyright (C) 1989 to 2009, 2010 BERTRAND Joel
>>>
>>> +++This is a free software with absolutely no warranty.
>>> +++For details, type 'warranty'.
>>>
>>> +++Fatal : Error in compilation
>>> dhp at uranus ~ $ "rpl" -v
>>> +++RPL/2 (R) version 4.0.14 (Wednesday 05/19/10, 14:23:18 CEST)
>>> +++Copyright (C) 1989 to 2009, 2010 BERTRAND Joel
>>>
>>>     Reverse Polish Lisp/2 version 4.0.14 for POSIX operating systems
>>>         Half-compiled, high-level procedural language,
>>>         mainly aiming at scientific calculations
>>>
>>>     Copyright (C) 1989, 1990, 1991, 1992, 1993, 1994, 1995, 1996, 1997,
>>> 1998,
>>>     1999, 2000, 2001, 2002, 2003, 2004, 2005, 2006, 2007, 2008, 2009, 2010
>>>     BERTRAND Joël
>>>
>>>     All rights reserved worldwide
>>>
>>> +++Error : Any executable definition
>>> dhp at uranus ~ $
>>
>>      Are you sure that rplpp is built with RPL/2 4.0.14 ? All external
>> programs are checksummed before execution. Typically, you try to launch
>> RPL/2 with an old rplpp (rplpp is RPL/2 preprocessor) or a rplpp that
>> was rebuilt since last rpl sucessfull build. If you have a log file of
>> your compilation, you can check that both sha1 and md5 sums are correct.
>
> In short ... i was using make install-strip. That caused a bug. When
> using make install and disabling all altervative way to strip (Gentoo
> tries to do it at two places) ... i get 4.0.14 working.
>
> So, either disable "make install-strip", or change your build/install
> process to make it homogenous/consistant :) Because, MD5 check bugs anyway.
>
> I found YAOB :)

	Today, don't use install-strip. I know this bug but I haven't time 
enough to fix it. install-strip target is added by automake and I have 
found only a workaround : allways do a strip -s on all checked tools. I 
haven't add this workaround to regular tarball because this workaround 
breaks RPL/C compiler.

	JKB