From: Kbwms AT aol DOT com Message-ID: <180.1de5394e.2c5bd182@aol.com> Date: Fri, 1 Aug 2003 10:21:54 EDT Subject: Re: Changes to Files in Archive djtst203.zip To: djgpp-workers AT delorie DOT com MIME-Version: 1.0 Content-Type: multipart/alternative; boundary="part1_180.1de5394e.2c5bd182_boundary" X-Mailer: 8.0 for Windows sub 6011 Reply-To: djgpp-workers AT delorie DOT com --part1_180.1de5394e.2c5bd182_boundary Content-Type: text/plain; charset="US-ASCII" Content-Transfer-Encoding: 7bit MartinS: No need to send separate email to me, I am on the list for djgpp-workers. In a message dated 8/1/2003 9:16:15 AM Eastern Standard Time, ams AT ludd DOT luth DOT se writes: > According to Kbwms AT aol DOT com: > > > > -- Start of included mail From: Kbwms AT aol DOT com > > > >Rich: > > > >The problems seems to be connected with math.h. Suggest you implement the > >makefile change only since the change to qhead.h in subdirectory > tgen/qfloat is > >purely cosmetic. Later, we can implement the change to qhead.h. > > > >There is a conflict between ieeefp.h and math.h that involves a definition > of > >isnanf(). See line 76 in ieeefp.h and line 149 in math.h. The means that > >the order of inclusion matters -- something that we are supposed to avoid > >altogether. > > When my C99 isnan macro has been added the function can be > removed. Its prototype can be changed to "#define isnanf(x) isnan(x)" > if math.h is included (which might not be optimal). > At the moment, my solution is to add a simple #if defined statement in the appropriate place. > Kbwms, you might not know that I've been hacking on the macros > fpclassify and is{nan,inf,normal,finite}. > Good for you. My attention is on the long double functions in C99, totally. I have code for fpclassifyl() that you are welcome to. KB Williams --part1_180.1de5394e.2c5bd182_boundary Content-Type: text/html; charset="US-ASCII" Content-Transfer-Encoding: quoted-printable MartinS:

No need to send separate email to me, I am on the list for djgpp-workers.
In a message dated 8/1/2003 9:16:15 AM Eastern Standard Time, ams AT ludd DOT luth.= se writes:

According to Kbwms AT aol DOT com:
>

-- Start of included mail From:  Kbwms AT aol DOT com


>Rich:
>
>The problems seems to be connected with math.h.  Suggest you implem= ent the
>makefile change only since the change to qhead.h in subdirectory tgen/qf= loat is
>purely cosmetic.  Later, we can implement the change to qhead.h. >
>There is a conflict between ieeefp.h and math.h that involves a definiti= on of
>isnanf().  See line 76 in ieeefp.h and line 149 in math.h.  Th= e means that
>the order of inclusion matters -- something that we are supposed to avoi= d
>altogether.

When my C99 isnan macro has been added the function can be
removed. Its prototype can be changed to "#define isnanf(x) isnan(x)"
if math.h is included (which might not be optimal).


At the moment, my solution is to add a simple #if defined statement in the a= ppropriate place.

Kbwms, you might not know that=20= I've been hacking on the macros
fpclassify and is{nan,inf,normal,finite}.


Good for you.  My attenti= on is on the long double functions in C99, totally.  I have code for fp= classifyl() that you are welcome to.


KB Williams --part1_180.1de5394e.2c5bd182_boundary--