Progress on cross-compilation
Re: Progress on cross-compilation
Yeah, you can delete the old one.
By working, I mean I have managed to compile something non-trivial (the file command), and upload that and test it on the NAS box to make sure it executes as expected
By working, I mean I have managed to compile something non-trivial (the file command), and upload that and test it on the NAS box to make sure it executes as expected
Hamish
Re: Progress on cross-compilation
So, some good news and some bad news.
The good news:
I've now built pretty much everything we wanted to build, plus some extras.
The bad news:
It seems GCC 4.7 and higher attempts to use features from kernel 3.1+ sometimes when compiling for arm < v7. The box runs Linux 2.6. So far, this has only happened with MariaDB, so I'm now building GCC 4.6 in the hope that that will fix it. Hopefully that will be that. If we encounter issues later, I may have to re-build everything, which will not be fun, but I will do it if needed
The good news:
I've now built pretty much everything we wanted to build, plus some extras.
The bad news:
It seems GCC 4.7 and higher attempts to use features from kernel 3.1+ sometimes when compiling for arm < v7. The box runs Linux 2.6. So far, this has only happened with MariaDB, so I'm now building GCC 4.6 in the hope that that will fix it. Hopefully that will be that. If we encounter issues later, I may have to re-build everything, which will not be fun, but I will do it if needed
Hamish
Re: Progress on cross-compilation
So, it turns out that it doesn't build with GCC 4.6, because it needs those features. I'm sensing a catch-22 coming on
Maybe it's time to give up and use the stock version of mysql. I will have another stab at it, and maybe ask the LUG tonight.
Maybe it's time to give up and use the stock version of mysql. I will have another stab at it, and maybe ask the LUG tonight.
Hamish
Re: Progress on cross-compilation
Okay, so seeing as this seems to be basically impossible to do without flash-upgrading the NAND on the NAS box (we need a newer kernel), I have decided not to build a newer version of the database server.
Hamish
Re: Progress on cross-compilation
Okay, I have successfully built and tested Gnuplot on the NAS box now. It seems to run well. The only catch is the built-in documentation isn't available, but I'm sure we can live without that.
Hamish
Re: Progress on cross-compilation
Matplotlib is now installed and tested working on the NAS box. It is slow, especially slow to load at first, but it does work and can export to JPG, PNG, and PDF.
It's probably too slow for custom plots to be created on demand, but for scheduled nightly plotting or similar it will do just fine. The NAS box also has ~500MB of swap space - I checked. This is of course easy enough to increase by tweaking my startup scripts if we find we need more.
It's probably too slow for custom plots to be created on demand, but for scheduled nightly plotting or similar it will do just fine. The NAS box also has ~500MB of swap space - I checked. This is of course easy enough to increase by tweaking my startup scripts if we find we need more.
Hamish