Active Topics
-
Camera phone competition March 2023 "Endless" (35)
to General by Maemish - 12 hrs, 23 mins ago -
TLS 1.2 for N9 is ready (28)
to MeeGo / Harmattan by smartblu9 - 13 hrs, 37 mins ago -
Maemo repository (3)
to Community by torsen - 1 day, 5 hrs ago -
Error after installing Leste on sdcard (11)
to Maemo 7 / Leste by Maemish - 2 days, 2 hrs ago -
Introducing ubiboot N9 (multiboot OS loader) (1,393)
to Alternatives by smatkovi - 3 days, 9 hrs ago - more...
| The Following User Says Thank You to Zeta For This Useful Post: | ||
|
2016-01-13
, 10:10
|
|
Posts: 2 |
Thanked: 7 times |
Joined on Nov 2009
|
#12
|
In fact, the main difference between Mer and Yocto is the way that packages are compiled.
Mer use scratchbox2, an emulator (VirtualBox) build packages natively for the target architecture.
Yocto use a cross-compiler to do this.
Some times ago, many projects don't cross-compile "mainline", so, the way "scratchbox" was easier to build a complete system (for example, 6 years ago, Xorg is very difficult to cross-compile).
Today, there is many ARM devices, and cross-compilation for any open source projects, is an important feature... New build systems (cmake, qmake, qbs...) managed cross compile, for autotool, there is some restrictions, but, Yocto provide recipe for many project (in order to patch autotool to accept cross-compiler, for example, by avoid to use some .c compiled at configure step, and execute to anaylse device features, or test build result).
So, I think that today, scratchbox2 is not a good way to generate a system. Yocto is more powerfull, more flexible, more open to contribution... It's easier to build new package, or to contribute by proposing a new layer on github for example.
Today, we can find many project to build some "exotic" packages.
We can find a layer on github, to build owfs-server, or eibd; it is very difficult if I want to build it into mer project.
A month ago, I want to try the head version of VLC on the Jolla Tablet (VLC now support wayland), but, after some hours of work, I abandoned...
To create a recipe for VLC on Yocto, there is lesser work... And in addition, we can find some layer with this recipe on the web.
If maemo was create with scratchbox is, from my point of view, historical, could have been a good (best) choice when the project started (2005).
Today, Yocto (OpenEmbedded) is a better technologie, I think it is a shame that Jolla use scratchbox rather than Yocto.
Last edited by condo4; 2016-01-13 at 10:48.
Mer use scratchbox2, an emulator (VirtualBox) build packages natively for the target architecture.
Yocto use a cross-compiler to do this.
Some times ago, many projects don't cross-compile "mainline", so, the way "scratchbox" was easier to build a complete system (for example, 6 years ago, Xorg is very difficult to cross-compile).
Today, there is many ARM devices, and cross-compilation for any open source projects, is an important feature... New build systems (cmake, qmake, qbs...) managed cross compile, for autotool, there is some restrictions, but, Yocto provide recipe for many project (in order to patch autotool to accept cross-compiler, for example, by avoid to use some .c compiled at configure step, and execute to anaylse device features, or test build result).
So, I think that today, scratchbox2 is not a good way to generate a system. Yocto is more powerfull, more flexible, more open to contribution... It's easier to build new package, or to contribute by proposing a new layer on github for example.
Today, we can find many project to build some "exotic" packages.
We can find a layer on github, to build owfs-server, or eibd; it is very difficult if I want to build it into mer project.
A month ago, I want to try the head version of VLC on the Jolla Tablet (VLC now support wayland), but, after some hours of work, I abandoned...
To create a recipe for VLC on Yocto, there is lesser work... And in addition, we can find some layer with this recipe on the web.
If maemo was create with scratchbox is, from my point of view, historical, could have been a good (best) choice when the project started (2005).
Today, Yocto (OpenEmbedded) is a better technologie, I think it is a shame that Jolla use scratchbox rather than Yocto.
Last edited by condo4; 2016-01-13 at 10:48.
| The Following 6 Users Say Thank You to condo4 For This Useful Post: | ||
|
|
2016-02-05
, 22:58
|
|
Posts: 339 |
Thanked: 1,623 times |
Joined on Oct 2013
@ France
|
#13
|
Hello condo4,
Sorry, I missed your reply earlier...
Thanks for the insight. I understand better now what would be the reasons to use one or the other.
Too bad there is no work in the yocto direction then. This would probably be a huge work, but in the same time, not doing it will probably result in huge work in other areas (like you mention).
Sorry, I missed your reply earlier...
Thanks for the insight. I understand better now what would be the reasons to use one or the other.
Too bad there is no work in the yocto direction then. This would probably be a huge work, but in the same time, not doing it will probably result in huge work in other areas (like you mention).
|
|
2016-02-06
, 11:58
|
|
Posts: 440 |
Thanked: 2,256 times |
Joined on Jul 2014
|
#14
|
FWIW, i think the Glacier guys are investigating switching to Yocto/OpenEmbedded for their build process rather than the Mer toolchain.
Asteroid OS is similar and already uses OE with nemo/mer packages as well as updated recipies from Qt and such.
Asteroid OS is similar and already uses OE with nemo/mer packages as well as updated recipies from Qt and such.
There are a lot of distributions out there, so it would be great to understand what Mer tries to do differently, how, and if it is still meaningful compared to other things that have come out since then.