some random crap coming from zodb *sometimes* #9412
Comments
|
comment:1
Robert Miller said: Should I add a dependency to deps to include setuptools in the then John Palmieri said: Looking at src/setup.py, I think that they seem to be a standard part |
|
comment:2
Robert, There are a couple of tickets open (#9274 and #9351 come to mind) to make some changes to 'deps', as a result of problems when setting SAGE_PARALLEL_SPKG_BUILD=yes. I think you need to balance how significant these bits of random crap are, against the potential for introducing problems if you make changes to 'deps'. IMHO, such changes are not low-risk, given the pivotal role that file plays. This is especially so if the changes are extensive, as they are in #9274. The problem with making major changes to 'deps' is it's easy to get it wrong and screw up things. That rather defeats your plan of having a decent stabilisation period, if you add quite risky changes in now. Hence you might want to consider 3 different possibilities.
There are some obvious problems with 'deps' now. For a start, most, if not all things in spkg/standard are not depending on things in spkg/base. So prereq in spkg/base, which checks the compilers and the build environment is generally sound, is not completing until after several standard packages in spkg/standard are built. On My Ultra 27, these are the order packages are built (ignore R and Maxima, as they are not built - I had to touch the relevant file in spkg/installed to fake these). The options to 'ls' show time stamps to the nearest nanosecond and are sorted in order of build. Note, since John Palmieri has made some updates to the documentation (#8263) which document SAGE_PARALLEL_SPKG_BUILD. That documentation specifically says SAGE_PARALLEL_SPKG_BUILD is experimental. I'm finding the SAGE_PARALLEL_SPKG_BUILD very good, and I can live with its minor problems though I'd like to see the problems resolved. It is one area of Sage able to benefit many people, so deserves higher priority than some of the more obscure Sage patches. Anyway, there is a few things for you to consider before adding any or all of these! Dave |
|
comment:3
Replying to @sagetrac-drkirkby:
Changes which cause problems can and will be reverted before the final release.
First, #9274 isn't even ready yet -- nothing is going in without a positive review. Second, we're not in the stabilization period anyway.
In fact I'm not doing any of these three choices.
This is entirely off topic for this ticket. The topic of this ticket is to fix one problem with deps, not rehaul it. Since you are aware of #9274, I have no idea why you're ranting here.
This is even further off topic! |
|
Attachment: deps.patch.gz |
|
Attachment: trac_9412-scripts.patch.gz |
|
Author: Robert Miller |
|
comment:5
Attachment: deps.gz |
|
comment:6
This looks good. setup.py does have: so clearly it needs setuptools. You don't have it depending on BASE - I assume that will be taken care of with #9274. Dave |
|
comment:7
Replying to @sagetrac-drkirkby:
Correct, that is the plan. Thank you for the review. |
|
Reviewer: David Kirkby |
|
Merged: sage-4.5.alpha3 |
In sage-4.5.alpha1, I see this:
As discussed in the thread
http://groups.google.com/group/sage-release/browse_thread/thread/7286e07fd3f06375
Component: build
Author: Robert Miller
Reviewer: David Kirkby
Merged: sage-4.5.alpha3
Issue created by migration from https://trac.sagemath.org/ticket/9412
The text was updated successfully, but these errors were encountered: