Re: [mongodb-dev] Roadmap for migration MongoDB, to s390x architecture

[Andy Schwerin]

All of the BSON encoding and decoding functionality assumes little-endian byte ordering.  By far, that will be the most complicated task.  The system was not built to be portable to other processor architectures, and preparing it for that will require extensive design work.  I wouldn't recommend doing that work at this time.  If you wanted to port to a little-endian system, your best route might be to use a C++11 compiler, and replace the existing types written in assembly with the appropriate C++11 std::atomic types and operations.  That would take you a long way.

On Thu, May 9, 2013 at 6:18 AM, Alberto Fernandez <scr...@gmail.com> wrote:

Hi all,
I would like to migrate MongoDB to s390x  architecture.  When I try to compile, I saw that this architecture is not supported.
Investigating a little more, I see that the problem is the assembler instructions, because s390x is big endian.

I've been taking a look to the source code and I found assembler code in several files.

How can I start to do this? I know that is not an easy task, but I only want to start, and see if i can have help.

Is there any similar project from another migration?

Do you think that is impossible?

Thanks in advance


Note: it seems to be more people interested:
http://www.mail-archive.com/linu...@vm.marist.edu/msg63124.html
https://groups.google.com/forum/#!msg/mongodb-dev/-Va-n3dV8y0/yDrri2tbeY4J

--
You received this message because you are subscribed to the Google Groups "mongodb-dev" group.
To unsubscribe from this group and stop receiving emails from it, send an email to mongodb-dev...@googlegroups.com.
To post to this group, send email to mongo...@googlegroups.com.
Visit this group at http://groups.google.com/group/mongodb-dev?hl=en-US.
For more options, visit https://groups.google.com/groups/opt_out.