Category Archives: soc

Notes on pkg_patch

My Google Summer of Code project for 2010 is "pkg_patch" - a utility and a framework for working with binary patches to FreeBSD binary packages. It is necessary because the default way of dealing with 3rd party applications / ports in most BSD systems is to build them from source via various frameworks like ports and pkgsrc, and so far the binary package side has been mostly neglected, at least in FreeBSD.

I strongly feel that the only thing that will help the adoption of binary packages (by making them useful) is to introduce the notion of "stable" ports but since I've repeatedly been told of the uselessness of banging on this particular drum, I'll leave it alone for others.

Anyway, most aspects of pkg_patch are described in the FreeBSD wiki, and these blog posts will deal with daily and miscellaneous information.

Read more...

Notes on pkg_patch

My Google Summer of Code project for 2010 is "pkg_patch" - a utility and a framework for working with binary patches to FreeBSD binary packages. It is necessary because the default way of dealing with 3rd party applications / ports in most BSD systems is to build them from source via various frameworks like ports and pkgsrc, and so far the binary package side has been mostly neglected, at least in FreeBSD.

I strongly feel that the only thing that will help the adoption of binary packages (by making them useful) is to introduce the notion of "stable" ports but since I've repeatedly been told of the uselessness of banging on this particular drum, I'll leave it alone for others.

Anyway, most aspects of pkg_patch are described in the FreeBSD wiki, and these blog posts will deal with daily and miscellaneous information.

Read more...

Back to the HEAD

Just when I thought I can finally finish a whole project on a RELENG branch without touching HEAD (and thus do it directly on my desktop), there it is - with a "D'Oh" level of surprise I find that the pkg_install infrastructure is, yes, different enough in HEAD and STABLE. Specifically, pkglib has been separated out and with it came the "pkg.h" header.

Which reminds me to be grateful about just how much finally having a decent virtualization environment - VirtualBox in this case - is worth! When I started doing FreeBSD development, also for a SoC project, i had a hard time working on it because running a debug kernel on the hardware I got at the time was too slow to have it on all the time and there were no quality virtualization products for FreeBSD. To top it all, I didn't have network connectivity at the time during the summer (when I returned home from the student dormitory) so using Perforce was really painful.

Read more...

Back to the HEAD

Just when I thought I can finally finish a whole project on a RELENG branch without touching HEAD (and thus do it directly on my desktop), there it is - with a "D'Oh" level of surprise I find that the pkg_install infrastructure is, yes, different enough in HEAD and STABLE. Specifically, pkglib has been separated out and with it came the "pkg.h" header.

Which reminds me to be grateful about just how much finally having a decent virtualization environment - VirtualBox in this case - is worth! When I started doing FreeBSD development, also for a SoC project, i had a hard time working on it because running a debug kernel on the hardware I got at the time was too slow to have it on all the time and there were no quality virtualization products for FreeBSD. To top it all, I didn't have network connectivity at the time during the summer (when I returned home from the student dormitory) so using Perforce was really painful.

Read more...