ABCDEFG
1
IDVotesType/MemberMilestonePriOwnerSummary/Comments
2
3
21330DEFECTtriageP4belleyIncorrect Dependency Cycle error
4
consensuswontfixSteven to open an enhancement for synthetic targets via Alias()
5
6
Bill
7
Brandon
8
David
9
GaryI'm on the fence. Running the target in AddPostAction is a common use case. What's the workaround if we don't change the behavior? There needs to be one. If we do change it, people can use Depends(), but will they think to? Probably not.
10
GregwontfixMy comment is in the issue. I don't think it's a bug. He can get the same effect with a synthetic target in a separate step.
11
Jim
12
Ken
13
Sohail
14
Steven1.0.xp4I think it's too common a use case to mark invalid as is. Right now, I'm thinking perhaps an option (yay, more configurability... :-/) that specifies whether or not the AddPostAction() command gets "scanned" for the implicit dependency. My gut say 1.0.x so the common use case gets looked at sooner rather than later, but p4 because in the grand scheme of things it's not killing anyone to have to specify Depends()...
15
16
21340FEATUREtriageP4stevenknightcreate public object like SCons.Util.CLVar
17
consensus
18
19
Bill
20
Brandon1.xIf it's cleaner to implement in python 2.x, I'd say wait for it.
21
David
22
Gary1.xp3Everyone should be using this rather than plain lists or strings, so I'm definitely in favor of it. Greg: what in python 2.x makes it easier? Seems like it works OK now. Maybe new-style classes make the syntax simpler?
23
GregGood idea, but when? Maybe it should wait until 2.0, when Python will have features to make this easier. (Gary: the 'attribute' machinery, which runs code when you assign to a varaible, so you could make it a CLVar for ever and ever.)

Generic comment: is there a list anywhere of all the names SCons adds to the global namespace? (Should there be?) It just seems to me that it's getting a bit cluttered in there, and maybe moving some of the names into clusters might be beneficial.
24
Jim
25
Ken
26
Sohail
27
Steven1.xp3stevenknightdefault reclassify-for-1.x settings
28
29
21350DEFECTtriageP4stevenknightSConscript(variant_dir) and VariantDir() inconsistant
30
consensus
31
32
Bill
33
Brandon1.0.x
34
David
35
Gary1.0.xp2ok w/ me.
36
Greg1.0.xp2One of mine. My comments are in the issue.
37
Jim
38
Ken
39
Sohail
40
Steven1.0.xp2agree w/greg
41
42
21360DEFECTtriageP4stevenknightsrc_dir with no variant_dir in SConscript() call
43
consensus
44
45
Bill
46
Brandon1.x
47
David
48
Gary1.xp2I'm OK with making this work.
49
Greg1.xp2One of mine. My comments are in the issue.
50
Jim
51
Ken
52
Sohail
53
Steven1.xp2agree w/greg.
54
55
21370DEFECTtriageP4stevenknightUser's Guide: rewrite VariantDir description
56
consensus
57
58
Bill
59
Brandon1.0
60
David
61
Gary1.0p3I love better doc, almost as much as better error messages.
62
Greg1.0p3One of mine. My comments are in the issue.
63
Jim
64
Ken
65
Sohail
66
Steven1.0p3agree w/greg
67
68
21380DEFECTtriageP4stevenknightMissing () in Entry.must_be_same
69
consensus
70
71
Bill
72
Brandon1.0.xp2
73
David
74
Gary1.0p2Pretty late to change anything for 1.0, but it looks clearly correct, and all tests pass with it on my Ubuntu box. So I vote for 1.0. I'd be even happier if there were a test case that failed with the code as is.
75
Greg1.0?If he's right, this should be fixed ASAP, if not sooner. What are the potential implications of puting this in 1.0? Does this require a 0.98.6? I agree w/Gary about a test case.
76
Jim
77
Ken
78
Sohail
79
Steven1.0.xp2?maybe p1? it's clearly the right fix, but i don't have a good feel for what all of the ramifications might be.
80
81
21400FEATUREtriageP4stevenknightAutomatic node re-ordering patch
82
consensus
83
84
Bill
85
Brandon1.x or 2.xFor a callback based mechanism I can see this going into 1.x, but for internal SCons functionality, I'd say 2.x
86
David
87
Gary
88
GregI'm staying out of this one; my reservations about it have been expressed in the mailing list. In brief, I can't see wasting any core developer time on a tiny niche like this, but if an absolutely perfect patch shows up, I won't resist applying it.
89
Jim
90
Ken
91
Sohail
92
Steven1.xp4After more thought, I'm okay with some clean hook that allows for re-ordering build order between targets, like the hook that --random uses allows within a target's dependencies, along with some configurability for how many targets get examined, and a way to fetch this info (e.g. at the end of the run) for external storage. I'm no longer in favor of putting this info in the .sconsign file. Someone wants to try to make something this smart? Fine, knock yourself out, but base SCons should just stay dumb and simple about it.
93
94
21410DEFECTtriageP4stevenknightg++ tool doesn't add -fPIC to SHCXXFLAGS as default case
95
consensus
96
97
Bill
98
Brandon1.0.x
99
David
100
Gary