gnupic: Re: Gpasm comments
Subject:
Re: FW: Gpasm comments
From:
Craig Franklin ####@####.####
Date:
20 Dec 2000 00:59:14 -0000
Message-Id: <00121919042800.00876@r2d2>
I responded to Simon's email yesterday not realizing that there is a limit
on the email size to the mailing list:
ezmlm-reject: fatal: Sorry, I don't accept messages larger than 30000 bytes (#5.2.3)
The patch and zip I attached pushed it over the limit. Oops.
Simon, did you recieve it OK? Since I mailed it directly to you, it should have
gone through.
After a good nights sleep I realized something. The patch I sent will make
Simon's test case work but it doesn't address the real issue. I believe that
when macro is defined (not called) all of the parameters in the definition
should be local. I ran a test case in MPASM and it behaves in this manner.
MPASM 02.61 Released TEST.ASM 12-19-2000 18:59:54 PAGE 1
LOC OBJECT CODE LINE SOURCE TEXT
VALUE
00001 #DEFINE NUM1 1
00002
00003 LIST P=16C84
00004
00005 FRED MACRO NUM1
00006 MOVLW NUM1
00007 ENDM
00008
00009 FRED NUM1
0000 3001 M MOVLW NUM1
00010 FRED 0X20
0001 3020 M MOVLW NUM1
00011
00012 END
Simon, I will fix the real problem and send it in as part of a canidate for
0.8.15. Until then, I think the case sensitive defines patch I sent will get
you by.
On Mon, 18 Dec 2000, you wrote:
>
> 1). I would just download the latest release at:
>
> http://www.dattalo.com/gnupic/gpasm-0.8.14.tar.gz
>
> 2). #defines are always case insensitive in GPASM 0.8.14. I have attached a
> patch to this email which allows case sensitive defines to be selected by the
> -c command line option. It is not throughly tested and may have a bug.
> However, it works with the test case in your email.
>
> To apply the patch do the following:
>
> 1. Place the file downloaded in answer 1). in an empty directory (/tmp).
> 2. Place a copy of the patch, case_sensitive_defines.patch, in the
> directory (/tmp).
> 3. gunzip gpasm-0.8.14.tar.gz
> 4. tar -xvf gpasm-0.8.14.tar
> 5. cp -R gpasm-0.8.14 gpasm
> 6. patch -p0 < case_sensitive_defines.patch
> 7. cd gpasm
> 8. ./configure
> 9. make
> 10. ./gpasm -v
>
> Let me know if it solves your #define problem. I will perform a more through
> test in the next couple of days. If it does the trick I will sent it in with
> other changes I have as a canidate for GPASM 0.8.15.
>
> 3). It is on my list. I was planning on taking care of it in a month or so if
> no one else tackles it first.
>
> 4.) The best thing to do is to extract the includes from the latest MPLAB
> distribution. Attached is the set from MPLAB v5.20.
>
> Keep in mind that some things they do in their includes are not
> yet supported by GPASM. Most of them are known and fixes have been
> identified. They should be in 0.8.15.
>
> On Mon, 18 Dec 2000, Simon Wood wrote:
> > It appears that James is no longer with SGI, so perhaps someone on this list can help with my comments/questions.
> >
> > Simon Wood
> >
> > > -----Original Message-----
> > > From: Simon Wood
> > > Sent: Monday, December 18, 2000 3:43 PM
> > > To: ####@####.####
> > > Subject: Gpasm comments
> > >
> > > Hi James,
> > > I've just started looking at PIC's and was glad to see that there are some Linux tools available. I downloaded the CVS version and am attempting to compile some code (to make a chorded keyboard - pickey.asm).
> > >
> > > Comments and questions:
> > > 1). The CVS download instructions on sourceforge neglect to tell you the module name (or is it an intelligence test??), I ended up using '.' which dumped everything in the local directory rather than creating a new one.
> > >
> > > 2). The code I'm using makes heavy use of macros, but your parser was falling over with 'Expression too complex'. After a bit of digging around I worked out it is because the parser is not case sensitive and it was getting confused by 'VALUE' and 'value' and seeing them as the same entity:
> > > #define VALUE 0x11
> > >
> > > fred macro value
> > > MOVLW value
> > > endm
> > >
> > > Is there an easy way to change this??
> > >
> > > 3). Again with macros, in the following the use of variable 'end' seems to confuse the parser.... doesn't really need to be fixed though, I'll just work around it with 'mend' or something.
> > > ensure_table: macro start, end
> > > if ((high start) - (high end)) !=0
> > > ERROR start
> > > endif
> > > endm
> > >
> > > 4). The code appears to use a 'standard' include file (P16C63.INC) for the common name/port specifications. I've had a look around on the web but couldn't find them. Do you have any pointers?
> > >
> > > Thanks in advance,
> > > Simon Wood
> > >
> >
> > ---------------------------------------------------------------------
> > To unsubscribe, e-mail: ####@####.####
> > For additional commands, e-mail: ####@####.####
----------------------------------------
Content-Type: text/x-c; name="case_sensitive_defines.patch"
Content-Transfer-Encoding: base64
Content-Description:
----------------------------------------
----------------------------------------
Content-Type: application/x-gzip; name="pic-include-5.20.tar.gz"
Content-Transfer-Encoding: base64
Content-Description:
----------------------------------------
