<?xml version="1.0" encoding="utf-8"?>
<rss version="2.0" xmlns:atom="http://www.w3.org/2005/Atom"><channel><title>Recent changes to feature-requests</title><link>https://sourceforge.net/p/prc-tools/feature-requests/</link><description>Recent changes to feature-requests</description><atom:link href="https://sourceforge.net/p/prc-tools/feature-requests/feed.rss" rel="self"/><language>en</language><lastBuildDate>Tue, 02 Dec 2003 20:32:03 -0000</lastBuildDate><atom:link href="https://sourceforge.net/p/prc-tools/feature-requests/feed.rss" rel="self" type="application/rss+xml"/><item><title>ARM Compiler support for __SPIC__</title><link>https://sourceforge.net/p/prc-tools/feature-requests/9/</link><description>&lt;div class="markdown_content"&gt;&lt;p&gt;Sample Code:&lt;br /&gt;
test.c:&lt;br /&gt;
#include &amp;amp;lt;PalmOS.h&amp;amp;gt;&lt;br /&gt;
// nothing else needs to be in file to reproduce&lt;/p&gt;
&lt;p&gt;[cygwin] c:/whatever$ arm-palmos-gcc test.c&lt;/p&gt;
&lt;p&gt;Results:&lt;br /&gt;
Code fails due to OS_CALL macro that evaluates to __SPIC__&lt;/p&gt;
&lt;p&gt;It's not a bug, because the compiler is working just&lt;br /&gt;
fine, but it prevents writing ARM code that calls into&lt;br /&gt;
PalmOS without thunking to 68k code running in&lt;br /&gt;
emulation.  For games or multimedia apps, performance&lt;br /&gt;
can be much higher if the code doesn't have to thunk.&lt;/p&gt;&lt;/div&gt;</description><dc:creator xmlns:dc="http://purl.org/dc/elements/1.1/">Dave Bacher</dc:creator><pubDate>Tue, 02 Dec 2003 20:32:03 -0000</pubDate><guid>https://sourceforge.net455c3731dfbc88a7d0bc920522b3aebdf18b2485</guid></item><item><title>Debugging over USB Cable</title><link>https://sourceforge.net/p/prc-tools/feature-requests/8/</link><description>&lt;div class="markdown_content"&gt;&lt;p&gt;Some devices as the M150 Zire do not have a serial port.&lt;/p&gt;
&lt;p&gt;It would be great to have the possibility to debug &lt;br /&gt;
applications on these devices using an USB connection.&lt;/p&gt;
&lt;p&gt;-Caspar&lt;br /&gt;
(seckendorff@alphatec.de)&lt;/p&gt;&lt;/div&gt;</description><dc:creator xmlns:dc="http://purl.org/dc/elements/1.1/">Anonymous</dc:creator><pubDate>Sat, 12 Apr 2003 11:56:12 -0000</pubDate><guid>https://sourceforge.net03c7125561e10d770049e2a62365d61fa1a23f3c</guid></item><item><title>build-prc: Indirect file input for arguments</title><link>https://sourceforge.net/p/prc-tools/feature-requests/7/</link><description>&lt;div class="markdown_content"&gt;&lt;p&gt;I have an application that has many, many small &lt;br /&gt;
resources.  (It's a reference manual)&lt;/p&gt;
&lt;p&gt;build-prc can't handle the number of PilRC .bin files on &lt;br /&gt;
the command line.  The total amount of data isn't &lt;br /&gt;
immense - it's just broken up into lots of little resources.&lt;/p&gt;
&lt;p&gt;If build-prc could take an @file syntax, in which &amp;amp;quot;file&amp;amp;quot; &lt;br /&gt;
contained the list of files to be bound (or even allowed &lt;br /&gt;
command line options to be included) this wouldn't be a &lt;br /&gt;
problem.  As it is, in order to be able to bind the .prc file, &lt;br /&gt;
I'm having to resort to combining resources in a way &lt;br /&gt;
that's significantly complicating my code.  (I'm trying to &lt;br /&gt;
avoid a separate database.)&lt;/p&gt;&lt;/div&gt;</description><dc:creator xmlns:dc="http://purl.org/dc/elements/1.1/">Kevin Hunter</dc:creator><pubDate>Fri, 31 Jan 2003 17:37:16 -0000</pubDate><guid>https://sourceforge.netaa8a25242f28e467e6b16fc8b2d15e8eb96e606c</guid></item><item><title>Support for gdb/mi on gdb.</title><link>https://sourceforge.net/p/prc-tools/feature-requests/6/</link><description>&lt;div class="markdown_content"&gt;&lt;p&gt;m68k-palmos-gdb don't recognise the -interpreter=mi &lt;br /&gt;
command line option, possibly because it as been &lt;br /&gt;
disabled.  I am building a gdb debugger interface for &lt;br /&gt;
palm based windows/cygwin environment and gdb/mi &lt;br /&gt;
support would be greatly appreciated.&lt;/p&gt;&lt;/div&gt;</description><dc:creator xmlns:dc="http://purl.org/dc/elements/1.1/">Steve Beaudoin</dc:creator><pubDate>Sat, 18 Jan 2003 17:55:21 -0000</pubDate><guid>https://sourceforge.netdcbb9e3052f33376d9a2fb68ca18472ba625bc3e</guid></item><item><title>preprocessor defines for SDK version</title><link>https://sourceforge.net/p/prc-tools/feature-requests/5/</link><description>&lt;div class="markdown_content"&gt;&lt;p&gt;With the usage of -palmos&amp;amp;lt;VERSION&amp;amp;gt; there should&lt;br /&gt;
be some way of determining what version of the SDK&lt;br /&gt;
is being used so that code which can handle multiple&lt;br /&gt;
versions can do so without the developer having to &lt;br /&gt;
make up their own defines to add to the compiler flags.&lt;/p&gt;
&lt;p&gt;i.e. &lt;/p&gt;
&lt;p&gt;echo foo | m68k-palmos-gcc -palmos3.5 -dM -E - &lt;/p&gt;
&lt;p&gt;and &lt;/p&gt;
&lt;p&gt;echo foo | m68k-palmos-gcc -palmos5 -dM -E - &lt;/p&gt;
&lt;p&gt;should have something in the output which is &lt;br /&gt;
different.&lt;/p&gt;
&lt;p&gt;Perhaps __palmsdk__ with a value of the version&lt;br /&gt;
multiplied by 10?&lt;/p&gt;&lt;/div&gt;</description><dc:creator xmlns:dc="http://purl.org/dc/elements/1.1/">Tim Hudson</dc:creator><pubDate>Sun, 05 Jan 2003 03:58:27 -0000</pubDate><guid>https://sourceforge.net15da6006f939faba55147adc782595caa43e17f8</guid></item><item><title>Automatic multi-segment applications.. implementation idea</title><link>https://sourceforge.net/p/prc-tools/feature-requests/4/</link><description>&lt;div class="markdown_content"&gt;&lt;p&gt;This is cut-and-paste from my two posts to&lt;br /&gt;
prc-tools-devel list to which I got no replies.&lt;/p&gt;
&lt;p&gt;First, regarding C++ code generation:&lt;br /&gt;
1. I have a larger project that uses exceptions. All&lt;br /&gt;
exception-related&lt;br /&gt;
data goes into the data segment. This makes my data&lt;br /&gt;
segment grow over 120kB!!&lt;/p&gt;
&lt;p&gt;2. What's worse, gcc generates exception data for all&lt;br /&gt;
functions, even those that don't catch or throw&lt;br /&gt;
exceptions (and I've verified in the assembler output&lt;br /&gt;
that in such cases these tables aren't even referenced).&lt;/p&gt;
&lt;p&gt;The solution: put exception-handling tables into the&lt;br /&gt;
same section as code, instead in the .data section.&lt;br /&gt;
This immediately follows with another feature request&lt;br /&gt;
(and solution idea) regarding limited code segment sizes:&lt;/p&gt;
&lt;p&gt;I've devised a way that could make automatic&lt;br /&gt;
generation of multi-segment applications easy, without&lt;br /&gt;
external tools and manual sectioning of the code.&lt;/p&gt;
&lt;p&gt;The approach would be this:&lt;br /&gt;
1. each compilation unit (.c file) goes into its own&lt;br /&gt;
section&lt;br /&gt;
2. when a subroutine X is about to be called:&lt;br /&gt;
- if the subroutine is in the same compilation unit,&lt;br /&gt;
normal code&lt;br /&gt;
(as is now) is generated; just a bsr X&lt;br /&gt;
- if the subroutine is in another compilation unit, the&lt;br /&gt;
compiler would generate indirect jsr via a lookup-table&lt;br /&gt;
(that can be in a segment of its own, not in the main&lt;br /&gt;
.data section). The pointer to that global lookup table&lt;br /&gt;
would be stored in a global variable (say&lt;br /&gt;
__$reloc_table) and some address register would be&lt;br /&gt;
loaded with it when the jump is about to take place (or&lt;br /&gt;
it can be dedicated just for that, if needed,  say %a3,&lt;br /&gt;
%a4 if no -mown-gp is specified or %a6 if&lt;br /&gt;
-fomit-frame-pointer is in effect). Then something like&lt;/p&gt;
&lt;p&gt;jsr X__$reloc(%a3)&lt;/p&gt;
&lt;p&gt;can be generated.&lt;/p&gt;
&lt;p&gt;3. Another tool would be written that would do the&lt;br /&gt;
following things:&lt;br /&gt;
- generate an assembly file with all relocation data&lt;br /&gt;
for all external&lt;br /&gt;
symbols, like this:&lt;/p&gt;
&lt;p&gt;.section .sreloc&lt;br /&gt;
.org 0&lt;/p&gt;
&lt;p&gt;__$reloc_table:&lt;br /&gt;
X__$reloc:&lt;br /&gt;
.long 0 # for subroutine X&lt;br /&gt;
Y__$reloc:&lt;br /&gt;
.long 0 # for subroutine Y&lt;/p&gt;
&lt;p&gt;where 0 would be replaced with the offset of X from the&lt;br /&gt;
beginning of its section. The tool could generate the&lt;br /&gt;
file with the help with&lt;br /&gt;
nm (or via libbfd library, but with nm is easier if&lt;br /&gt;
using perl to write this tool)  The tool would then&lt;br /&gt;
call the assembler to assemble the relocation table,&lt;br /&gt;
construct appropriate linker scripts and call the linker.&lt;/p&gt;
&lt;p&gt;4. At run-time the startup code would check out all&lt;br /&gt;
code segments, and the relocation segment and apply&lt;br /&gt;
appropriate relocations (high 2 bytes of relocation in&lt;br /&gt;
reloc_table can be the segment number, while low 2&lt;br /&gt;
bytes can be the offset from the beginning of the&lt;br /&gt;
segment; the new table would be constructed at startup&lt;br /&gt;
and copied into new record when all base addresses of&lt;br /&gt;
all segments are known).&lt;/p&gt;
&lt;p&gt;THE CATCH: I know how to do everything described above&lt;br /&gt;
(including writing the necessary startup code), but I&lt;br /&gt;
don't know how do modify the code generation in the&lt;br /&gt;
compiler. So if anybody is willing to give me a hand&lt;br /&gt;
here, I'd do the rest of the job.&lt;/p&gt;
&lt;p&gt;====== &lt;/p&gt;
&lt;p&gt;Or is there any reason you people DON'T WANT to&lt;br /&gt;
implement automatic multi-segment applications and&lt;br /&gt;
usable C++ code generation? I can't believe that people&lt;br /&gt;
who hack gcc can't devise such a simple idea and&lt;br /&gt;
implement manual sectioning of the code instead... I&lt;br /&gt;
don't want this to sound as accusation, but a friend of&lt;br /&gt;
mine suggested that you may even have a deal with&lt;br /&gt;
Metrowerks so that they can continue to sell their&lt;br /&gt;
compiler...&lt;/p&gt;&lt;/div&gt;</description><pubDate>Fri, 27 Sep 2002 05:39:38 -0000</pubDate><guid>https://sourceforge.net93d4d4db24736cd2783414e67413d9b5947df277</guid></item><item><title>Palm OS 5</title><link>https://sourceforge.net/p/prc-tools/feature-requests/3/</link><description>&lt;div class="markdown_content"&gt;&lt;p&gt;Will PRC-TOOLS support developing OS 5 applications &lt;br /&gt;
with ARM native codes?&lt;/p&gt;&lt;/div&gt;</description><dc:creator xmlns:dc="http://purl.org/dc/elements/1.1/">Tony Yat-Tung Cheung</dc:creator><pubDate>Thu, 15 Aug 2002 16:11:22 -0000</pubDate><guid>https://sourceforge.net7c61edc631a12e6c6b71d1a261a3345070b955c7</guid></item><item><title>PRC-Tools binary for OSX?</title><link>https://sourceforge.net/p/prc-tools/feature-requests/2/</link><description>&lt;div class="markdown_content"&gt;&lt;p&gt;Since the PRC-Tools are UNIX based, are there any plans&lt;br /&gt;
to release compiled binaries for OSX? AFAIK this would&lt;br /&gt;
be the first noncommercial compiler for Palm appls for&lt;br /&gt;
ODX...&lt;/p&gt;
&lt;p&gt;Thanks,&lt;br /&gt;
Marc&lt;/p&gt;&lt;/div&gt;</description><dc:creator xmlns:dc="http://purl.org/dc/elements/1.1/">Marc Mennigmann</dc:creator><pubDate>Thu, 11 Jul 2002 21:16:34 -0000</pubDate><guid>https://sourceforge.net0924bd3e5c39b279e205922e8b8896934c095fb7</guid></item><item><title>Further documentation on .def files</title><link>https://sourceforge.net/p/prc-tools/feature-requests/1/</link><description>&lt;div class="markdown_content"&gt;&lt;p&gt;I think it would be wonderful to get more &lt;br /&gt;
information on what a .def file can contian!  I &lt;br /&gt;
couldn't find this in any of the documents I &lt;br /&gt;
received with the source or on the webpage.&lt;/p&gt;
&lt;p&gt;I am aware of the&lt;br /&gt;
application { &amp;amp;quot;App Name&amp;amp;quot; CRID options }&lt;br /&gt;
and&lt;br /&gt;
multiple code { &amp;amp;quot;seg1&amp;amp;quot; &amp;amp;quot;seg2&amp;amp;quot; }&lt;br /&gt;
but am curious if any other things can be &lt;br /&gt;
controlled.&lt;/p&gt;&lt;/div&gt;</description><dc:creator xmlns:dc="http://purl.org/dc/elements/1.1/">Jim Ramsay</dc:creator><pubDate>Thu, 11 Apr 2002 04:56:16 -0000</pubDate><guid>https://sourceforge.netcd30cca95f50420816d19bd0dc1f9ea43b67715f</guid></item></channel></rss>