2010-05-06 19:36:05     Duplicate development environments

Document created by Aaronwu Employee on Aug 21, 2013
Version 1Show Document
  • View in full screen mode

2010-05-06 19:36:05     Duplicate development environments

Reggy Perrin (UNITED STATES)

Message: 89206   

 

Hi folks,

 

We have 3 different developers working on an application, and need to synchronize their development environments (logistically, we can't use the same build machine for all 3).

 

We are having a difficult time keeping these systems in sync.  For instance, one team is developing the kernel & root file system, another the final application, and another diagnostics.

 

What's a strategy for keeping these 3 systems compatible?  We started with the same base, now seem to be having compatibility issues with libc and some other core libraries (the application requires building more packages than the kernel, and we are trying to manage when/where to put those additional libraries on our system).

 

Ideally, I'd love a method for extracting the core configuration from the kernel/rfs build environment, and propagating that to the other two environments.  Then, they could add the additional libraries to the configuration and build.

 

My gut instinct is that most developments utilize a single build environment, and don't see this as an issue.  Some time Googling has not been revealing (or I'm expressing this in incorrect search terms).

 

Any advise?

 

RP

QuoteReplyEditDelete

 

 

2010-05-06 20:49:50     Re: Duplicate development environments

Robin Getz (UNITED STATES)

Message: 89207   

 

Reggy:

 

Most people use some sort of source code management system.

 

We use svn/git. The config files are just managed inside the dist.

 

https://docs.blackfin.uclinux.org/doku.php?id=managing_kernel_upgrades#adding_your_board_to_the_uclinux_distribution

 

When someone needs to make a change to the config (add a lib or something), they just tell people to "Update Default Vendor Settings"

 

As for the toolchain - installing the toolchain on a server on /opt, and having people all mount that ro over the network - as long as everyone is on the same type of machine (Linux 32-bit) - it works pretty painless.

 

-Robin

Attachments

    Outcomes