Tips for all developers using the CVS

[Speedy2206]

I just wanted to make sure everyone knew about a few tips and tricks with the CVS, in regards to making changes and compiling.

Some people are under the impression that to make changes once a compile has failed you have to start again, which can take several hours. Why do this to yourself, when the build was only 10 minutes away from completion??

A few shortcuts that I know of:

ONCE YOU HAVE COMPILED AN IMAGE, YOU CAN STILL MAKE CHANGES AND COMPILE THOSE WITHOUT HAVING TO START FROM SCRATCH

SIMPLY ADD --enable-maintainer-mode TO THE END OF YOUR CONFIGURE COMMAND


Examples:
If you want to add code (maybe for testing, or as a last minute extra feature that you forgot to include) to the:

Linux Kernel

rm .deps/linuxkernel or rm .deps/linuxdir
make linuxkernel
make flash-neutrino-jffs2-all

Neutrino Binary

rm .deps/neutrino
rm ../apps/tuxbox/neutrino/src/.libs/neutrino
at this point if you want to add a new menu for example, modify MAKEFILE.AM to include the new .cpp file, and write the code.
make neutrino
make flash-neutrino-jffs2-all

Zapit Binary

rm .deps/zapit
rm ../apps/dvb/zapit/src/.libs/zapit
make any changes to ZAPIT at this point
make zapit
make flash-neutrino-jffs2-all

You can also do this for Enigma or any other plugin, etc. Each folder usually contains a .libs subdirectory which houses the main binary file for that component. Simply deleting it will force the environment to compile it again, including any changes you've made.

-----------------------------------------------------------------------------------------------

My personal way of making quick changes to neutrino without having to start from scratch or flash a new image.

For this technique you will need a Neutrino image already on your DBOX. Also you will need to add the path to your compiler.

Lets say you downloaded the CVS to /home/me/myimage. If your configure line was ./configure --prefix=/home/me/myimage/output --with-cvsdir=/home/me/myimage then I would write the following

export PATH="$PATH:/home/me/myimage/output/cdk/bin"

(this only has to be done once per session).
Basically you take the path from the --prefix part of the configure line, and add /cdk/bin to the end of it.

This basically just gives us quick access to the compiler and strip functions.

OK, moving on from that.

Lets say I built neutrino once, and that image is flashed on my DBOX already. But I want to add something else to the neutrino binary without starting again.

I would do:

cd ../apps/tuxbox/neutrino/src
... at this point I would make any changes to the source code, including Makefile.am (you should never adjust Makefile itself, Makefile.am will rebuild Makefile correctly if it detects changes)
rm .libs/neutrino
make
cd .libs
powerpc-tuxbox-linux-gnu-strip neutrino
cp neutrino /home/me

Afterwards I will find a neutrino file in my home directory, which I can FTP onto my DBOX. On the DBOX itself, I can telnet to quickly see the changes:

killall -9 start_neutrino
killall -9 neutrino
chmod 755 /tmp/neutrino
/tmp/neutrino &
...keep the telnet window open... closing it will kill /tmp/neutrino!...

To sum up...

If the original image I made was great, but was missing the (for example) Zapit Config menu that renwich has just completed. I could add that by modifying /home/me/myimage/apps/tuxbox/neutrino/src/gui/Makefile.am to include my zapitconf.cpp file. I could then copy the zapitconf.cpp and zapitconf.h files into the gui directory, and add a link to that menu within neutrino_menu.cpp.

I then run the make command again, which detects and compiles my changes, and gives me a neutrino binary file (in src/.libs) which I can FTP to my DBOX2 to see my new changes. I move the neutrino file to /home/me because any folder with a . before it is hidden by default, and putting it in my home directory also gives me quick and easy access to it.


Voila!
Hope that helps somebody

Any other tips to speed things up to prevent you having to do a whole re-compile from scratch just ask me and I'll probably have the answer. For each image I do I probably only download the CVS once, patch and modify as much as I like, then flash to box to test. If it all works I take the raw neutrino-jffs2.img2x and upload it here.

But lets just say that my image would have probably been flashed only once on my DBOX and then individual binary files FTP'd to it later for testing


- Speedy

[Speedy2206]

Also if you make alot of command-line programs like I do (makefta for example), by doing the following trick.....

export PATH="$PATH:/home/me/myimage/output/cdk/bin"

.....you can also quickly compile your program from anywhere, even outside of the cvs tree.

Simply type (for example):

powerpc-tuxbox-linux-gnu-g++ makefta.c -o makefta
powerpc-tuxbox-linux-gnu-strip makefta

And your binary file is ready to be used on the DBOX2 - all you need to do is FTP it and CHMOD 755. Done.

[Speedy2206]

Similarly I talked about shrinking files (see here: http://www.ukcvs.org/Forums/showthread.php?t=12379)

To achieve those results I done the following.
Anyone can do this if they have a linux box and have previously compiled a whole image. Even if the image is old and you have made new directories containing more up-to-date CVS trees, you can still use this trick.

Simply grab a file out of an existing image that you want to shrink (I would suggest trying /lib/modules/2.4.36.6/fs/cifs/cifs.o as a starting point), copy it to your linux machine (I chose a folder called /home/me/shrink), and type:

export PATH="$PATH:/home/me/myimage/output/cdk/bin"
cd /home/me/shrink
powerpc-tuxbox-linux-gnu-strip cifs.o

Afterwards look at the file size. Shrink, if it has any effect on the file you are trying it with, will usually halve the size of a file with no memory impact.

[Speedy2206]

And finally, if you find that you're using this trick alot (like me), it might be worth copying the files to a seperate directory, for example:

mkdir /home/me/dbox2compile
cd /home/me/myimage/output/cdk/bin
cp -r * /home/me/dbox2compile

Your export line will then look like this:

export PATH="$PATH:/home/me/myimage/dbox2compile"

After that you can delete ALL CVS trees from your computer and still have the compiler and strip functions ready for use.

Remember, the EXPORT line only needs to be typed once per session. So unless you shut down or reboot your machine, you should only type it once.

[pt-1]

Originally Posted by Speedy2206

Similarly I talked about shrinking files (see here: http://www.ukcvs.org/Forums/showthread.php?t=12379)

To achieve those results I done the following.
Anyone can do this if they have a linux box and have previously compiled a whole image. Even if the image is old and you have made new directories containing more up-to-date CVS trees, you can still use this trick.

Simply grab a file out of an existing image that you want to shrink (I would suggest trying /lib/modules/2.4.36.6/fs/cifs/cifs.o as a starting point), copy it to your linux machine (I chose a folder called /home/me/shrink), and type:

export PATH="$PATH:/home/me/myimage/output/cdk/bin"
cd /home/me/shrink
powerpc-tuxbox-linux-gnu-strip cifs.o

Afterwards look at the file size. Shrink, if it has any effect on the file you are trying it with, will usually halve the size of a file with no memory impact.

Sorry to highjack this but a user has reported problems with scanning/epg after he used shrink on the sectionsd so not every binary can be shrinked !

[PaphosAL]

Originally Posted by pt-1

Sorry to highjack this but a user has reported problems with scanning/epg after he used shrink on the sectionsd so not every binary can be shrinked !

T-C have got FreddyFrog's sectionsd down to 115kb now, pt-1, working 100% fine and no useage of /tmp whatsoever, m8!

Cheers- AL

Should have added- this was not with shrink!