Experimental Physics and
| |||||||||||||||||
|
On Friday, August 20, 2010, Ralph Lange wrote:On the MSI move: I am volunteering to move it into dbtools, keeping the history. (I have been reluctantly doing too many repository conversions lately, and this one is actually interesting: it will require merging from an unrelated repo, which probably makes "swiss army knife" git the best candidate for the operation. I will create a bzr branch on LP that you can merge in after 3.14.12 is out.)The problem with moving msi into base is that it makes it more difficult to apply site-specific changes. This may not be a strong argument against, though. dbLoadTemplate (ioc based) is part of base, the rules for host-side template instantiation are in base, dbLoadTemplate (host based) used to be part of base, and when msi was created as a replacement, I think it was a wrong decision to unbundle it. The generic functionality of msi (as well as the use cases, and implementing dbLoadTemplate functionality on the host as a "spec") are clear - I think we can provide a reasonable implementation that avoids the need for local patches. The idea is to keep the MSI variable in the build system (just pointing to the base version as default - instead of relying on a msi binary being in the user's PATH), so using a site-specific version will be just as easy as it is now. And - last not least: moving msi into base increases the code visibility and makes it easier to improve the shared version (instead of keeping improvements site-specific). BTW No "lists" involved here. In this case core-talk would have been appropriate for following-up, true. Any ideas how to make phone calls and discussions on the hallway public and archived? ;-) Cheers, Ralph
| ||||||||||||||||
ANJ, 02 Sep 2010 |
·
Home
·
News
·
About
·
Base
·
Modules
·
Extensions
·
Distributions
·
Download
·
· Search · EPICS V4 · IRMIS · Talk · Bugs · Documents · Links · Licensing · |