May 1999 - Gpc

Re: bash
by Nick Burrett 11 May '99

11 May '99

"Russ Whitaker" <russwhit(a)mind.net> writes: > hi frank > > i'm not a bash expert either: it took me 2 hours to come up with the > following > > enter the following 6 lines of code in a file called test > > #!/bin/sh > if [ $1 = $2 ]; then > echo "ok" > else > echo "oops" > fi > > next, chmod 777 test > then try it: > ./test 1 1 > and > ./test 1 2 > > now go back and change the = to == and try it and you get > > ./test: [: ==: binary operator expected > oops I tried this with bash 2.02 and there is no problem since = and == are interpreted the same in this case. Are you sure you are executing the script with `bash' and not `sh' ? The first line would seem to suggest `sh' which may be the cause of the problem. Nick.

1 0

bash
by Russ Whitaker 11 May '99

11 May '99

hi frank i'm not a bash expert either: it took me 2 hours to come up with the following enter the following 6 lines of code in a file called test #!/bin/sh if [ $1 = $2 ]; then echo "ok" else echo "oops" fi next, chmod 777 test then try it: ./test 1 1 and ./test 1 2 now go back and change the = to == and try it and you get ./test: [: ==: binary operator expected oops note: bash is from slakware 3.5 thanks russ

1 0

Re: Fw: gcc-19990330
by Frank Heckenbach 10 May '99

10 May '99

Once upon a time ago, Russ Whitaker wrote: > > near the end of make LANGUAGES=pascal there is a number of "if" > > statemnets followed by the message > > > > /bin/sh: [: ==: binary operator expected Unless this has already been solved, can you give more information where this accurs (perhaps a log of the make run)? > > p.s. think "Trim" should trim both spaces and tab chrs > > ( now only spaces) low priority wish list What does the standard say about this? (`Trim' is a predefined function of the standard, so we have to care about the standard here.) Technically, of course, it would be easy to do it. Alternatively (if the standard wants only spaces), we could also make a different routine for spaces and tabs... Frank -- Frank Heckenbach, frank(a)fjf.gnu.de, http://fjf.gnu.de/ GPC To-Do list, latest features, fixed bugs: http://agnes.dida.physik.uni-essen.de/~gnu-pascal/todo.html

2 1

Re: Fw: gcc-19990330
by Frank Heckenbach 10 May '99

10 May '99

Russ Whitaker wrote: > symptom: GPC would compile unit crtc.c but wouldn't link to ncurses > > so used text editor to search the Makefile.in files for == and replace with > = when it appeared in an "if" statement > then did cnfigure, etc, etc > > result: gpc now finds ncurses > > (bash defaults to always false when it sees == ) Sorry, I don't think I can follow you... First of all, I'm not so much of a bash expert. What do you mean by "bash defaults to always false when it sees =="? My man page says `==' may be used in place of `=' in comparisons, and I can't seem to find any difference in bash's behaviour. Does it mean that (some version of ?) bash does not recognize `==' as a comparison operator? Secondly, I'm a bit surprised about the symptom. The only `==' I could find were one in Make-lang.in in the main GPC source directory, and two in Makefile.in in the rts subdirectory. All of them were of the form "if [ x$(WITH_SHARED) == xyes ] ; then ...", to test whether one wants to build a shared library of the RTS. I can't see any connection to not linking ncurses. -- Well, perhaps, due to the make failing here, it did not get to the point of installing the new units, so GPC used the units of an older version which did not refer to ncurses correctly -- that might be the case... Anyway, replacing these three `==' by `=' (which I did now) fixes the whole problem, right? (Or just this symptom?) Frank -- Frank Heckenbach, frank(a)fjf.gnu.de, http://fjf.gnu.de/ GPC To-Do list, latest features, fixed bugs: http://agnes.dida.physik.uni-essen.de/~gnu-pascal/todo.html

1 0

Re: Linking problems
by Frank Heckenbach 10 May '99

10 May '99

Peter Gerwinski wrote: > When I `ar' together a bulk of `.o' files compiled with GPC, > I get lots of undefined references between the `.o's in the `.a'. Did you `ranlib' the library? Though I couldn't seem to produce any problems without doing so with my sources, in general that's necessary, AFAIK... > The same compilation with `--autobuild' works. What do you mean? Do you compile the files with `--autobuild', then `ar' together the resulting objects, or do you use `--autobuild' without `ar'? What happens if you just link the object files? Does it make any difference whether they were compiled with `--autobuild' or `--automake'? Frank -- Frank Heckenbach, frank(a)fjf.gnu.de, http://fjf.gnu.de/ GPC To-Do list, latest features, fixed bugs: http://agnes.dida.physik.uni-essen.de/~gnu-pascal/todo.html

1 0

Re: Fw: gcc-19990330
by Russ Whitaker 09 May '99

09 May '99

Hi Frank symptom: GPC would compile unit crtc.c but wouldn't link to ncurses so used text editor to search the Makefile.in files for == and replace with = when it appeared in an "if" statement then did cnfigure, etc, etc result: gpc now finds ncurses (bash defaults to always false when it sees == ) ---------- > From: Frank Heckenbach <frank(a)tim.gerwinski.de> > To: gpc(a)gnu.de > Subject: Re: Fw: gcc-19990330 > Date: Sunday, May 09, 1999 7:02 AM > > Once upon a time ago, Russ Whitaker wrote: > > > > near the end of make LANGUAGES=pascal there is a number of "if" > > > statemnets followed by the message > > > > > > /bin/sh: [: ==: binary operator expected > > Unless this has already been solved, can you give more information > where this accurs (perhaps a log of the make run)?

1 0

Global variables problem
by Ian Thurlbeck 07 May '99

07 May '99

Dear GPC-ers It's a while since a reported a bug in gpc, but hopefully this one is just me being stupid (again). I have a main program and a library program which share some global variables and I want to set the variable in the main program and print them out from the library program: ------main.p--------- program main(input,output); #include "main.h" begin global_int:=100; global_str:='Goodbye World'; print_globals; end. -------------------- ------lib.p----------- module lib(input,output); #include "main.h" var global_int:integer; global_str:string(40); procedure print_globals; begin writeln('global_int=',global_int); writeln('global_str=',global_str); end; end. ------------------- ------main.h-------- var global_int: external integer; global_str: external string(40); procedure print_globals; external; ------------------- Fairly straightforward, but I get: ida globals/ % gpc -o main main.p lib.p ida globals/ % main global_int=100 global_str= Not time to say goodbye yet, it seems. If I move the declarations of the global variable from lib.p to main.p I get: ida globals/ % main global_int=100 global_str=Goodbye World Success, but why ?? Many thanks for a great compiler Ian -- Ian Thurlbeck http://www.stams.strath.ac.uk/ Statistics and Modelling Science, University of Strathclyde Livingstone Tower, 26 Richmond Street, Glasgow, UK, G1 1XH Tel: +44 (0)141 548 3667 Fax: +44 (0)141 552 2079

1 0

Re: Interested about GNU Pascal development
by Nick Burrett 07 May '99

07 May '99

Peter Gerwinski <peter(a)gerwinski.de> writes: > Hello, folks! > > > GrimReaperX wrote: In fact, I am willing to do whatever I can to help > > Linux/Unix*/GNU to grow.... > > > > I agrre with your position, And I'm happy beause I've just found the > > GNU-Pascal. I'm french and I have problems with my english, but if you do > > some translation ant need somebody to help you (only in french), I'm here! > > Having French translations of GNU Pascal documentation is > certainly a Good Thing. What would probably be the best thing to do is internationalise the Pascal compiler with GNU gettext and then translate all the warning and error messages into various languages. > Just one thing to take into account: The GNU Pascal > documentation is currently fairly incomplete. If you find > a missing section, be welcome to add it. Translating all the GNU Pascal documentation into another language can create a maintenance headache as you have to try and keep track of the alterations that will be made to the document. Certainly a one-off translation would probably be good, but trying to keep it up-to-date is likely to be more trouble than it is worth. Nick.

1 0

Re: Problem compiling regexdemo and pipedemo on Redhat linux 5.2
by Frank Heckenbach 05 May '99

05 May '99

David James wrote: > > BTW, I think one should generally *not* use `Reply-To' headers in > > the mailing list, unless in special cases (like the strawpoll > > recently), just like one should not ask for private replies in > > newsgroups. > > My apologies, my MUA puts in a reply-to header by default, I've never > needed to remove it on any other mailing list, but looking at mail > from those most of them appear to re-write the reply-to header to be > the mailing list submission address. > > With my MUA, if I try to reply to your message it is going to go > direct to you and not to the list. So perhaps this list should insert a reply-to-the-list header be default? Peter, can the mailing list software do that? Is that desirable? Frank -- Frank Heckenbach, frank(a)fjf.gnu.de, http://fjf.gnu.de/ GPC To-Do list, latest features, fixed bugs: http://agnes.dida.physik.uni-essen.de/~gnu-pascal/todo.html

2 1

Re: Problem compiling regexdemo and pipedemo on Redhat linux 5.2
by david＠tcs01.demon.co.uk 05 May '99

05 May '99

In message <7DB86F9E.19990505201803.FOO-2516.frank(a)goedel.fjf.gnu.de> Frank Heckenbach writes: > I answered this mail yesterday, but then I saw that the original > posting had a `Reply-To' header, so I'm sending it again, to the > list, because it might be of interest to others. > > BTW, I think one should generally *not* use `Reply-To' headers in > the mailing list, unless in special cases (like the strawpoll > recently), just like one should not ask for private replies in > newsgroups. My apologies, my MUA puts in a reply-to header by default, I've never needed to remove it on any other mailing list, but looking at mail from those most of them appear to re-write the reply-to header to be the mailing list submission address. With my MUA, if I try to reply to your message it is going to go direct to you and not to the list. > > David James wrote: > > > /usr/lib/gcc-lib/i586-pc-linux-gnu/2.8.1/units/regexc.c:24: rxposix.h: No > > such file or directory > > For the RegEx unit, you need the rx library. Sources and binaries > are available at > ftp://agnes.dida.physik.uni-essen.de/gnu-pascal/libs/ Thank you, I've now picked that up. Am I likely to need any of the other items from the libs directory? I ask because I see there is a gmp in there, but the demos using gmp work fine on my system ... > > > /usr/lib/gcc-lib/i586-pc-linux-gnu/2.8.1/units/pipec.c:34: > > `pid_t' undeclared (first use in this function) > > Oh yes, this is a bug. Please add the line > > #include <sys/types.h> > > in pipec.c (near the other #include statements), then it should > work. Thanks again. -- David James mailto:david@tcs01.demon.co.uk

1 0