Mike Frysinger wrote:
>>> Due to shell portability issues (which is to say shell scripts are not
>>> portable -- think Windows), convert elf2flt to C code.
Actually I did this a year ago - however, only for the subset
of the functionality our company internally needs, so it is not
really ready for publication.
>> Honestly my gut reaction to this is "yuck". Not your implementation, just
>> the idea in general. C is great for lots of things, but string-heavy
>> control flow driving external utilities surely isn't one of them.
>
> i agree, but facing the lack of any saner alternatives, here we are. i'm
> certainly open to different ideas.
Perl would be another alternative. From my experience the most of the
portability issues comes from shell differing in handling escaping
and variable expansion and some come from sed variants. Both functionality
is pretty standardized in perl and there are easily available
free or free-as-a-beer windows versions.
>> Maybe I'm biased because I also have a pretty lukewarm feeling about doing
>> embedded Linux development in a Windows environment. Seems like a great
>> way to ensure a mediocre output from a product team, but anyway...
Erm, strong words... A few real-life examples:
1) Our company develops a cross-platform product targetting Windows,
Windows CE, ucLinux and some real-time operating systems. Nothing
beats Visual Studio in productivity (for affordable price)
for developing the system-independent parts (which is 95%).
2) Some of our customers want to link their code with our libraries
(in fact, we are required to enable this, as there is some
LGPL and similarly licensed code). I can't force them
to install Linux (nor do I want to support them). I can't
even force them to install DLL-hell-prone cygwin - fortunately
there is the MinGW project.
3) ucLinux is maybe only a part of the solution. If a product team
is doing 90% developing for Windows, there is no sane reason
to switch to Linux.
> while i agree, my opinion doesnt matter terribly much. the customers are
> paying the bills and the customers get what they're going to pay for. they
> arent going to pay for "let's ditch windows completely".
Exactly.
>> If I was to vote for (or had time for) serious labour investment in elf2flt
>> it would be a cleanup and refactor of the horrible "do_relocs" function in
>> elf2flt itself - that thing is torture.
I'd vote for integrating into binutils... what elf2flt does is a horrible
hack, e.g. breaking any sane package management software ("where is your
real ld today?").
Regards
--
Stano
_______________________________________________
uClinux-dev mailing list
uClinux-dev@uclinux.org
http://mailman.uclinux.org/mailman/listinfo/uclinux-dev
This message was resent by uclinux-dev@uclinux.org
To unsubscribe see:
http://mailman.uclinux.org/mailman/options/uclinux-dev
