Message: 5
Date: Tue, 7 Dec 2010 15:29:27 +0000
From: Keith Marshall <>
Subject: Re: [Mingw-users] No FlexLexer.h
Message-ID: <>
Content-Type: text/plain;  charset="iso-8859-1"

On Tuesday 07 December 2010 07:11:02 Charles Wilson wrote:
On 12/6/2010 8:39 PM, Simson Garfinkel wrote:
I never use the library, as I always provide my own main() and
__yywrap. ?But I do need the FlexLexer.h.

Then you must be developing a C++ scanner.  Flex supports that,
certainly, but in a manner branded very visibly as experimental.  The
core purpose of flex is to generate C scanners, and these don't need
FlexLexer.h at all.

Yep. I'm developing a C++ scanner. (Or, rather, I developed one.)  It turns out that flex isn't too insistent about which FlexLexer.h you use, so I'm just using one from another platform and it works.

With respect, if you want to play with experimental (alpha) features,
then you really should have sufficient "nouse" to provision the header
for yourself, (and the library, if you need it).

In fact, I am doing that. 

main() != __main()

This is true, but it *is* main() and yywrap() which are provided
by -lfl.  The OP may be clouding the issue, by reference to __yywrap;
it is his own yywrap() which he should provide, (or he should use
the %option noyywrap feature to obviate the necessity).

My big recommendation is that the current strategy of having the
MinGW native compiler ignoring /usr/ is probably a mistake.

No, it most certainly is not. ?

Agreed.  I've already pointed the OP to the relevant MinGWiki docs,
which explain why MinGW's GCC doesn't look in /usr -- what *is* /usr
anyway, on a bare MS-Windows box?  He isn't going to win any sympathy
by continuing to beat this drum.

I believe that I stopped beating this drum when the library issue was explained.