is cygwin port supported?
Dmitry Semikin
Ralf Hemmecke
> /home/semikin/builds/fricas_110929/fricas-build/build/i686-pc-cygwin/bin/notangle:
> No such file or directory
looks somehow suspicious. You obviously don't have noweb installed.
Noweb is only in the distributed fricas packages but not in the source
tree, you if you compile from trunk you have to make sure that noweb is
otherwise installed.
Ralf
Dmitry Semikin
looks somehow suspicious. You obviously don't have noweb installed. Noweb is only in the distributed fricas packages but not in the source tree, you if you compile from trunk you have to make sure that noweb is otherwise installed./home/semikin/builds/fricas_110929/fricas-build/build/i686-pc-cygwin/bin/notangle:
No such file or directory
Dmitry Semikin
Dmitry Semikin
here is several words about this:
Waldek Hebisch
allowed to define symbols not mentioned in the standard and I could
not see 'getline' in the standard. And via '-ansi' we request strict
ANSI compliance, so all extensions should be deactivated.
Anyway, there was no change to nowed and it did compile on Cygwin,
so that must be relatively recent Cygwin change.
However, given broken Cygwin 'stdio.h' it seems that the only solution
is to rename getline in noweb to new name, say 'noweb_getline'.
--
Waldek Hebisch
heb...@math.uni.wroc.pl
Gabriel Dos Reis
[...]
| However, given broken Cygwin 'stdio.h' it seems that the only solution
| is to rename getline in noweb to new name, say 'noweb_getline'.
OpenAxiom got rid of noweb a while ago. Do you still need it?
-- Gaby
Waldek Hebisch
Yes, noweb is used to unwrap content of algebra, etc and input
subdirectories.
--
Waldek Hebisch
heb...@math.uni.wroc.pl
Gabriel Dos Reis
| >
| > Waldek Hebisch <heb...@math.uni.wroc.pl> writes:
| >
| > [...]
| >
| > | However, given broken Cygwin 'stdio.h' it seems that the only solution
| > | is to rename getline in noweb to new name, say 'noweb_getline'.
| >
| > OpenAxiom got rid of noweb a while ago. Do you still need it?
| >
|
| Yes, noweb is used to unwrap content of algebra, etc and input
| subdirectories.
Yes, I understand. What I meant is that OpenAxiom developed its own
tangling/untangling tool to make these build issues and dependencies
evaporate.
-- Gaby
Dmitry Semikin
Dmitry Semikin
Dmitry Semikin
Ralf Hemmecke
https://github.com/hemmecke/prolitex
Then .pamphlet would really be .tex (thanks to the powerful listings
package). The format change would be from
<<NAME>>=
code here
@
to
\begin{chunk}{NAME}
code here
\end{chunk}
and an option to select if you want <<...>> or any other delimiters for
a reference to another chunk inside the code chunk.
weaving is not necessary anymore.
tangling is done by a little awk script.
I was, however, not yet eager enough to propose such a change to the
fricas sources.
Ralf
Ralf Hemmecke
No. I tried to start adding automake to fricas where all the standard
targets would be for free, but it's not that easy, because the build is
not so standard and automake rather supports .c compilation better than
.boot or .lisp. So you'll have to wait a bit.
However, if you look into the "install" target in the top-level
Makefile.in, you'll find that you can do a uninstall manually quite easily.
First, find your previous install path.
Look at the top of the fricas script. There is a variable AXIOM.
It will point to the installed files of FriCAS.
It ends in something similar to this...
..../lib/fricas/target/x86_64-unknown-linux
Remove the whole fricas directory.
Remove there the fricas and efricas scripts. From the bin directory
(which should be at the same level as the lib dir).
That's it.
Ralf
Dmitry Semikin
Dmitry Semikin
Waldek Hebisch
Use
make all-input
to run the tests. It should take about 15-30 minutes and produce
about 2.5 Mb of results. The results are printed to standard output,
but also go to files (src/input/*.output in the build directory).
Some tests print statistics about results but for the other (most)
one has to compare the results with results from other runs.
--
Waldek Hebisch
heb...@math.uni.wroc.pl
Dmitry Semikin
Waldek Hebisch
>
> --20cf303bffd208055f04ae58d0f5
> Content-Type: text/plain; charset=KOI8-R
> Content-Transfer-Encoding: quoted-printable
>
> Once more about building trunk version under cygwin.
>
> I've tried to it one more time and finished with following situation:
>
> Build failed with the error:
>
> >> System error:
> LOAD: A file with name
> RLIB/IEVALAB
> does not exist
>
> directory
> make[3]: *** [stamp-db] =EF=DB=C9=C2=CB=C1 1
> make[3]: Leaving directory
> `/home/dmitry_semikin/builds/fricas-trunk/fricas-build/src/algebra'
>
>
> I looked at that directory and found, that there is IEVALAB.erlib directory
> and IEVALAB-.NRLIB (with dash at the end). So, for some reason compiler
> added dash at the end of IEVALAB, while creating .NRLIB directory and all
> files within it.
>
IEVALAB-.NRLIB is OK. Spad compiler should create both IEVALAB-.NRLIB
and IEVALAB.NRLIB.
> I tried to look into the source spad files, but did not find there any
> differences from other spad files (the only thing I've noticed is that afte=
> r
> )abbrev line there is '--' comment (second line in file)... but I do not
> think, it can play so (although I did not try to build without this line to
> make sure, that it is not important)
>
> So, I wonder if sombody has any idea about why this can happen (dash '-'
> sign is appended to the name of .erlib, when converted to .NRLIB)? Or may b=
> e
> somebody can give some hints about how this can be debugged... I'll try to
> think about it myself, but I'm not quite familiar with fricas building
> process and compiler internals yet.
>
One possibility is to capture all messages to a file (for sucessful
full build expect about 18 Mb of messages). After failure search
for messages about compilation of IEVALAB.spad. If there is an
error message it may explain what is happening. If this does not
help there is possibility to run compiler by hand. The failing stage
can be re-run by going to 'src/algebra' subdirectory, running
(note setting environment variables, otherwise interpsys will
not start)
AXIOM=./axp DAASE=./r1 ../../build/arch/bin/interpsys
where 'arch' is your build architecture (probably i386-pc-cygwin)
and typing
)set break break
)read "boo_db.input"
to the interpsys prompt. Note: ')set break break' is so that
on error you will get into Lisp debugger. Unfortunately,
the failing stage is doing a lot of work that has to be done
sequentlialy in single run.
--
Waldek Hebisch
heb...@math.uni.wroc.pl
Dmitry Semikin
Waldek Hebisch
FYI, I regularly test with clisp on Linux. The last time I tested
with clisp-2.48 on 64-bit Debian 6.0.
--
Waldek Hebisch
heb...@math.uni.wroc.pl