2of1

half a nibble of another

Some fantastic news to report!

Remember this issue?

Well, today I received an email from Mark Lawrence at Microsoft in regard to the issue of using the Windows SDK headers in FOSS.

Here’s an excerpt of the relevant bit:

"Creating a derivative of the d2d1.h header file would be a problem,
but merely looking at it for referential purposes in order to develop 
your own header and software for the Windows platform should be 
within the scope of the SDK license."

So it seems we have the go-ahead for ‘peeking’ in the original headers and using them as a base reference for our own!

Seg7 is a simple library for control of 7-segment displays using the Arduino (or any compatible clone of course).
Read the rest of this entry »

I’ve been looking for a way to compile an app developed with Qt Creator on Linux for Windows.

I came across this but I needed to tweak it before it worked so I decided to blog my own mini-guide (tested working with with Qt 4.6.3).

This assumes that you already have the Qt SDK (for Linux) installed.

1. The first thing to do is download and install the MinGW cross compiler
2. Then download the Windows version of the Qt SDK. As it’s distributed as a Win package, you’ll need to either install it on windows and copy the necessary files across to your linux distro (what I did) or install it through Wine (not sure if this even works). You can cross-compile it as well if you’re being adventurous
3. Place the Qt Win SDK libraries in /usr/lib32/qt4win32 and headers in /usr/include/qt4win32
4. Copy /usr/share/qt4/mkspecs/win32-g++/ to /usr/share/qt4/win32-x-g++/
5. In the above directory, replace qmake.conf with the one found here
6. Make sure that the paths in qmake.conf match those of your system (see NOTE below); specifically QMAKE_CXX, QMAKE_LINK, QMAKE_INCDIR, QMAKE_INCDIR_QT, QMAKE_LIBDIR_QT
7. Run qmake adding –spec win32-x-g++ command line arg. This should cross-compile your app.

NOTE: As my linux box is a x86_64, my /usr/lib is /usr/lib32. Please adjust this accordingly for your setup.

There’s a quick and effective way of producing a random value in C if you only require it at most once per second using current time as a random seed:

#include <time.h>     // time_t, time()
#include <stdlib.h>   // srand(), rand()
 
time_t s;
time(&s);
srand((unsigned int)s);
 
// calc random value (limits are inclusive)
int rand_val = 1 + (int)((float)UPPER_LIMIT * 
               (rand() / (RAND_MAX + (float)LOWER_LIMIT)));

Of course reading from /dev/urandom is good too (on linux).

There’s a special command to do this:

.tables

Sometimes it’s useful to ’squash’ commit; in essence merging them into a single commit.

This can be done using the interactive mode of rebase:

git rebase -i origin

The commits can then be change from pick to either squash or fixup (squash that discards commit’s log message).

The order of the commits can be changed as well, so for example:

pick ea4f69f DVB: Increased signal timeout times
pick d58bac8 DVB access module format improvements
pick 5461a01 DVB-S scanning support
pick 1a5a61a DVB channel search dialog now shows ETA
pick 351bcd2 small syntax fix

Say I want to squash 351bcd2 into 5461a01 and discard the log message, I simply change it as follows:

pick ea4f69f DVB: Increased signal timeout times
pick d58bac8 DVB access module format improvements
pick 5461a01 DVB-S scanning support
fixup 351bcd2 small syntax fix
pick 1a5a61a DVB channel search dialog now shows ETA

The easiest way to undo a rebase is to find the relevant commit as it was immediately before the rebase using git reflog.

You can then reset to that commit; e.g.:

git reset --hard HEAD@{50}

I’m hoping to improve DVB support in VLC over the next few months.
Read the rest of this entry »

All uncommitted changes:

git reset --hard

Specific uncommitted changes:

git checkout <file(s)>