Building newRPL from source code
Preparing a system for compiling the simulator
Linux
The following instructions are for Ubuntu, package names might change slightly on other distros.
Packages to install (either use synaptic package manager, or apt):
git libelf-dev libudev-dev
If using other distros, verify that these other packages are installed:
gcc g++ libc6-dev
and for Qt, some 3D related headers (even though not used by newRPL) are needed, therefore these packages need to be installed as well:
libgl-dev
Time to install the Qt framework.
- Go to http://qt-project.org and download the open source version. It will download a file named
qt-unified-…
; - change the permissions to allow executing it as a program and run it to install Qt;
- when it's time to select components, it's very important to open the lists and select a library version and target. The installer will not select a library version by default, and failing to select one will leave you without a proper library and unable to build newRPL. Select the latest Qt version (5.10.0 at the time of this writing), the only component needed is the one labeled
Desktop gcc 64-bits
. Make sure at least that component is checked.
freeBSD
In order to use Qt Creator, a graphical desktop needs to be installed (that's outside the scope of this article).
Packages to install (either using pkg or OctoPkg):
git qt5 qtcreator
And that's all that's required.
For developers wishing to cooperate with the project, another package is needed:
gdb
Once the packages are installed,
- start Qt Creator, and navigate the menus and dialogs through
Tools
,Options…
,Build&Run
,Kits
and select theDesktop
(default) configuration; - verify that everything is populated, and change the Debugger to
System GDB at /usr/local/bin/gdb
which is the version of gdb that was just installed. This is needed only to debug the code, not needed if the objective is simply to have a running version of newRPL Desktop.
Preparing a system for cross-compiling the firmware
Linux
The following instructions are for Ubuntu, package names might change slightly on other distros.
Packages to install (either use synaptic package manager, or apt):
gcc-arm-none-eabi
Other packages like
binutils-arm-none-eabi
will be installed automatically, on other distros this might be different and it needs to be installed separately.
freeBSD
Packages to install (either use pkg or OctoPkg):
arm-none-eabi-gcc
Getting the source code
Linux and freeBSD
While at your user's home, open a terminal and run:
git clone https://git.code.sf.net/p/newrpl/sources newrpl-sources
Building the tools
elf2rom
The first tool that needs building is elf2rom
.
- Open Qt Creator, select
Open Project
, and go intonewrpl-sources/tools/elf2rom
; - select the
elf2rom.pro
project file; - when prompted, just click on
Configure project
; - at the bottom-left of the screen, there will be an icon that looks like a screen and shows
Debug
: change it toRelease
; - on the left toolbox, select the wrench
Projects
to access the build settings; - make sure
Release
is the selected configuration, thenAdd Build Step
/Make
, and in theMake arguments:
box typeinstall
; - now from the top menu,
Build
,Run QMake
, thenRebuild All
; - from the bottom of the screen, check the
Compile Output
pane to see if everything went well. Quite often, the last stepmake install
says there's nothing new to install when there's actually a newly built executable. Another round ofRun QMake
andRebuild All
usually shows in the output pane that the file is being copied tonewrpl/tools-bin
which is the expected result.
It's done: the elf2rom
tool was built and installed in the appropriate location to be used by the other projects. Select Close all projects and editors
from the main menu to end the session.
bmp2font
- Open Qt Creator, select
Open Project
, and go intonewrpl-sources/tools/fonts/bmp2font
; - select the
bmp2font.pro
project file; - when prompted, just click on
Configure project
; - at the bottom-left of the screen, there will be an icon that looks like a screen and shows
Debug
: change it toRelease
; - on the left toolbox, select the wrench
Projects
to access the build settings; - make sure
Release
is the selected configuration, thenAdd Build Step
/Make
, and in theMake arguments:
box typeinstall
; - now from the top menu,
Build
,Run QMake
, thenRebuild All
; - from the bottom of the screen, check the
Compile Output
pane to see if everything went well. Quite often, the last stepmake install
says there's nothing new to install when there's actually a newly built executable. Another round ofRun QMake
andRebuild All
usually shows in the output pane that the file is being copied tonewrpl/tools-bin
which is the expected result.
It's done: the bmp2font
tool was built and installed in the appropriate location to be used by the other projects. Select Close all projects and editors
from the main menu to end the session.
newRPL standalone compiler
- Open Qt Creator, select
Open Project
, and go intonewrpl-sources
; - select the
newrpl-comp.pro
project file; - when prompted, just click on
Configure project
; - at the bottom-left of the screen, there will be an icon that looks like a screen and shows
Debug
: change it toRelease
; - on the left toolbox, select the wrench
Projects
to access the build settings; - make sure
Release
is the selected configuration, thenAdd Build Step
/Make
, and in theMake arguments:
box typeinstall
; - now from the top menu,
Build
,Run QMake
, thenRebuild All
; - from the bottom of the screen, check the
Compile Output
pane to see if everything went well. Quite often, the last stepmake install
says there's nothing new to install when there's actually a newly built executable. Another round ofRun QMake
andRebuild All
usually shows in the output pane that the file is being copied tonewrpl/tools-bin
which is the expected result.
It's done: the newrpl-comp
tool was built and installed in the appropriate location to be used by the other projects. Select Close all projects and editors
from the main menu to end the session.
Building the newRPL desktop simulator
- Open Qt Creator, select
Open Project
, and go intonewrpl-sources
; - select the
newrpl-ui.pro
project file; - when prompted, just click on
Configure project
; - at the bottom-left of the screen, there will be an icon that looks like a screen and shows
Debug
: change it toRelease
; - on the left toolbox, select the wrench
Projects
to access the build settings; - make sure
Release
is the selected configuration, thenAdd Build Step
/Make
, and in theMake arguments:
box typeinstall
; - now from the top menu,
Build
,Run QMake
, thenRebuild All
; - from the bottom of the screen, check the
Compile Output
pane to see if everything went well. Quite often, the last stepmake install
says there's nothing new to install when there's actually a newly built executable. Another round ofRun QMake
andRebuild All
usually shows in the output pane that the file is being copied tonewrpl-sources
which is the expected result.
It's done: the newrpl-ui
tool was built and installed in the appropriate location to be used by the other projects. Select Close all projects and editors
from the main menu to end the session.
At the lower left area of the screen, use the upper Play button (Run without debugging) to start the application.
Building the firmware
- Open Qt Creator, select
Open Project
, and go intonewrpl-sources
, then select the relevant project file:newrpl-fw.pro
for the 50g target,newrpl-fw39.pro
for the 39g+ / 39gs target,newrpl-fw40.pro
for the 40gs target; - when prompted, just click on
Configure project
; - open a Terminal window and type
arm-none-eabi-gcc -print-libgcc-file-name
It will print the full path to a file called
libgcc.a
. Remember that path (without the/libgcc.a
part); - back to Qt Creator, from the left pane open the file
newrpl-fw.pro
, and scroll down to a line that readsGCC_LIBDIR = xxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxx
Change that path to match exactly the path we got from the previous step;
- one last check: locate the line that reads
INCLUDEPATH += firmware/include newrpl xxxxxxxxxxxxxxx
The
xxxxxxxxxxxxxxx
should point to suitable headers. In particular on a 64-bit system the system headers might be incompatible and cause errors. The first attempt should be to keep the defaults, and if any errors appear, to change it to/usr/include/newlib
(or wherever there's a suitable 32-bit library). While newRPL will not link to or use the library in any way, some standard headers are expected to provide definitions; - another possible alternative for 64-bit systems is to install the package
libc6-dev-i386
which in most cases should allow to keep the default directories. For Ubuntu it works with the defaults, for other systems change that to
/usr/include/newlib
, so the line readsINCLUDEPATH += firmware/include newrpl /usr/include/newlib
- save and close the file;
- now from the top menu,
Build
,Run QMake
, thenRebuild All
; - from the bottom of the screen, check the
Compile Output
pane to see if everything went well. There might be some warnings, but there should be no hard errors. There could be many warnings, depending on your version of gcc and your system: these warnings change all the time but they are harmless.
To verify that it completed the build, the last output line should read
Binary image written to “newrplfw.bin”
(for HP50g target) or
Binary image written to “newrpl39.bin”
(for HP39g+ / 39gs target) or
Binary image written to “newrpl40.bin”
(for HP40gs target)
If you see that line, everything went well, and your build directory (usually named build-newrpl-fw-Desktop…
) will contain the firmware file, which is the ROM that needs to be flashed on the calculator.
Updating the source tree to a new version
When updates are made to the source tree
- open a terminal session and go inside the main source directory (usually
newrpl-sources
), and issuegit pull
This will update the source code to the lastest version;
- build the standalone compiler and check the compile output to make sure the new version was copied to
newrpl/tools-bin
, if not then repeat the stepsRun QMake
andRebuild all
; - with the updated compiler in place, close all projects and rebuild the desktop simulator. The
Run QMake
step is not optional, it needs to be manually executed to make sure the build number is updated, otherwise theVERSION
command will report the incorrect build number; - finally, close all projects and rebuild the firmware.