Your message dated Sat, 06 Sep 2008 13:42:27 -0500
with message-id <[EMAIL PROTECTED]>
and subject line Resolution from upstream
has caused the Debian Bug report #291160,
regarding flex no longer lex compatible
to be marked as done.
This means that you claim that the problem has been dealt with.
If this is not the case it is now your responsibility to reopen the
Bug report if necessary, and/or fix the problem forthwith.
(NB: If you are a system administrator and have no idea what this
message is talking about, this may indicate a serious mail system
misconfiguration somewhere. Please contact [EMAIL PROTECTED]
immediately.)
--
291160: http://bugs.debian.org/cgi-bin/bugreport.cgi?bug=291160
Debian Bug Tracking System
Contact [EMAIL PROTECTED] with problems
--- Begin Message ---
Package: flex
Version: 2.5.31-26
The version of flex in Debian testing no longer works with a
large number of packages which depend on traditional lex
compatibility or POSIX lex compliance. The biggest problem
I have had is that yywrap() now takes an argument of some
type.
Because of this I don't think the /usr/bin/lex, /usr/lib/libl.a,
and the /usr/share/man/man1/lex.1.gz compatibility symlinks
are appropriate anymore.
There is a second issue, which is that the info and man
documentation fail to describe the compatibility issues. It
would seem to be appropriate to document when the yywrap()
function is discussed as well as in sections talking about
porting "AT&T" code (which is of course not the only code which
will have problems anymore -- POSIX and BSD code will also
fail). Similarly, the section talking about POSIX compliance
says:
"`flex' is fully compliant with the POSIX `lex' specification,
except that when using `%pointer' (the default), a call to
`unput()' destroys the contents of `yytext', which is counter
to the POSIX specification."
I believe that is an incorrect and misleading statement.
I am happy to see, now that I am submitting this bug report,
that there is a blurb in the package description and a
reference to flex-old. Maybe that should be the default
version of flex to be installed.
Finally, I'm happy to break this into multiple reports if it
is deemed appropriate.
Thanks,
-Ross
--- End Message ---
--- Begin Message ---
tags 291160 +upstream, wontfix
thanks
Hi,
Upstream has rejected the premise of this bug report, and at
this point, whether or not flex remains POSIX compliant is moot.
What matters is that flex is not going to change course, and people
needing a POSIX compliant lexer can start off from flex-old as a base.
manoj
John Said:
> Manoj Said:
>> John43said:
>> POSIX says nothing about yytext_ptr.
> Quite correct, POSIX says nothing about yytext_ptr.
> But it does speak about unput; and unput is defined like so:
> #define unput(c) yyunput( c, (yytext_ptr) )
> Also, look at the definition of yyless (again, required by
> POSIX), and note that it has YY_DO_BEFORE_ACTION:
> /* Return all but the first "n" matched characters back to
> the input stream. */
> #define yyless(n) \
> do \
> { \
> /* Undo effects of setting up
> test_ser_nr_Cemtext. */ \
> int yyless_macro_arg = (n); \
> YY_LESS_LINENO(yyless_macro_arg);\
> *yy_cp = (yy_hold_char); \
> YY_RESTORE_YY_MORE_OFFSET \
> (yy_c_buf_p) = yy_cp = yy_bp +
> yyless_macro_arg - YY_MORE_ADJ; \
> YY_DO_BEFORE_ACTION; /* set up
> test_ser_nr_Cemtext again */ \
> } \
> while ( 0 )
> And again note that YY_DO_BEFORE_ACTION is defined thus
> (taken out of the tests directory):
> #define YY_DO_BEFORE_ACTION \
> (yytext_ptr) = yy_bp; \
> test_ser_nr_Cemleng = (size_t) (yy_cp - yy_bp); \
> (yy_hold_char) = *yy_cp; \
> *yy_cp = '\0'; \
> (yy_c_buf_p) = yy_cp;
> When yytext_ptr is undefined, these call would fail, and
> thus the POSIX violation.
--
"To steal from one person is theft. To steal from many is taxation."
Daiell's Law (a take-off on Felson's Law)
Manoj Srivastava <[EMAIL PROTECTED]> <http://www.golden-gryphon.com/>
1024D/BF24424C print 4966 F272 D093 B493 410B 924B 21BA DABB BF24 424C
--- End Message ---
