develooper Front page | perl.perl5.porters | Postings from December 2001

perlio: still unfunctional

Thread Next
Ilya Zakharevich
December 19, 2001 23:38
perlio: still unfunctional
Message ID:
I tested the build with perlio on OS/2 again, and it is still too
fragile to be useful.  [And the test suite is useless: it detects none
of the problems listed below.]

0)  I do not even discuss the extreme unefficiency of the
    implementation.  [Last time I checked, it would not even export
    the buffer if in NL-translation mode.]

b)  perl -we "print qq(\n\r)" | perl -we \
	 "while(<>){ print qq(\n); printf qq(%%#x ), ord for split //}"
    dies with
	 ptr wrong 4b804 != 4b803 fl=00404500 nl=0 e=4b805 for 1 at -e line 1.

c)  The bug I reported half a year ago with a complete explanation
    what breaks and where and how to fix it is still there:

    when reading in text mode one-key-at-a-time one needs to duplicate
    C-j: the first one of the pair is lost.

d)  Embed.t prints "ok N" messages, but some of them are terminated by
    "\n\r" (as they should), some of them by "\n" (marked /*BIN*/
    below).  Embed.t is doing the following:

    PerlInterpreter *my_perl = perl_alloc();
    char *cmds[] = { "perl","-e", "print qq[ok 5\\n]", NULL };

#define my_puts(a) if(puts(a) < 0) exit(666)
    my_puts("ok 2");
    my_puts("ok 3");
    perl_parse(my_perl, NULL, (sizeof(cmds)/sizeof(char *))-1, cmds, env);
    my_puts("ok 4");					/*BIN*/
    my_puts("ok 6");					/*BIN*/
    my_puts("ok 7");                                    /*BIN*/
    my_puts("ok 8");                                    /*BIN*/

("ok 5" is printed in normal way.)  Hmm, this was done judging by the
screen occurence.  If I redirect to the file, then "ok 2" and "ok 3"
are also output with just "\n".

e) Due to a complete failure of "b" I cannot even think of any other test.

Hope this helps,

Thread Next Perl Programming lists via nntp and http.
Comments to Ask Bjørn Hansen at | Group listing | About