Account Options>

  1. Sign in
Issues 2008-09-01
 Share
The version of the browser you are using is no longer supported. Please upgrade to a supported browser.Dismiss

 
$
%
123
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
|
 
Still loading...
ABCDEFG
1
IDVotesType/MemberMilestonePriOwnerSummary/Comments
2
3
21310DEFECTtriageP2stevenknightenv.Glob causes spurious rebuilds
4
consensus
5
6
Bill
7
Brandon
8
David
9
Garycode 1.1GregI think it should sort. What's the good of not sorting it? It's trivial to do once in scons rather than making every user do it, and giving them obscure failure modes like this bug when it happens. Ordering function doesn't matter as long as it's deterministic from run to run (and within a run of course).
10
Gregdoc 1.1meI still say that Glob() should not sort, no more than glob.glob() does. And it's not obvious what the ordering function should be: the basename or the full path are both logical candidates under different circumstances. Add something to the manual to say that if youi want it deterministic, sort it yourself; it's one line of code. Whatever the decision (doc or code), give it to me for 1.1 and I'll fix it; I have some unfinished business with Glob().
11
Jim
12
Ken
13
Sohail
14
Stevencode 1.1GregGood build semantics should take precedence over Python semantics. The specific sort order "doesn't matter" so long as it's consistent from build to build.
15
16
13070DEFECTtriageP3pankratovereager get_contents() caching
17
consensus
18
19
Bill
20
Brandon
21
David
22
Garypatch for 1.1 (any time)LudwigAgree w/ Greg. It's quite good as is, but there's more to be done. I recommend making a new bug ticket for the remaining issue and trying to understand it as well as possible before designing a solution (if any).
23
Gregapply patch; discuss APIThe patch should be applied (although 'move' should clear both the target and the source caches). The API change needs discussion on the mailing list; this is not the right forum for deciding it.
24
Jim
25
Ken
26
Sohail
27
Stevenpatch for 1.1Ludwigagree w/Greg
28
29
19730ENHANCEMENTtriageP3stevenknightRemove src_dir unless someone can explain it
30
consensus
31
32
Bill
33
Brandon
34
David
35
Gary1.xgregGreg, you just want to remove the doc for src_dir, right? Per your last diff in the ticket (via email)? Seems OK to me I guess. Don't actually make it stop working in the code, cause some people do use it (I do). What about this case:
for d in subdirs:
....SConscript(d+"/SConscript", build_dir=get_build_dir(d), src_dir=d)
That's an idiom I use successfully.
36
Greg1.xp1Reopened. See my comments in the issue.
37
Jim
38
Ken
39
Sohail
40
Steven1.xp1steveknightI'll take another stab at it. I thought I had explained its usefulness (execute like the SConscript file was in the src_dir), but I can also see Greg's point that there are a lot of corner cases
41
42
20870DEFECTtriageP3issues@sconsJar builder cannot handle local classes
43
consensus
44
45
Bill
46
Brandon
47
David
48
Gary1.xp3Ludwigagree w/ others. Hard problem in general because needs to know about internals of Java compilers
49
Greg1.xp3Ludwig? Mati? David? other?Was originally 1.x p3; needs to be assigned to someone else.
50
Jim
51
Ken
52
Ludwig1.xmeAnonymous classes need a number in the file name. Should be fixed in the Java emitter (parse_java in JavaCommon). However, this will be fragile as it depends on how Java emits anon class files (seems to be implementation specific).
53
Mati
54
Sohail
55
Steven1.xp3LudwigThe real difficult part is that there might be different behaviors in different Java versions. (there definitely are for anonymous classes.) Any solution here needs to be validated against any/all common Java versions (1.[456], IIRC )
56
57
21830PATCHtriageP4stevenknightadding .sx to ASPP automatic actions
58
consensus
59
60
Bill
61
Brandon
62
David
63
Gary1.0.xno strong opinion. Ideally we should make it much easier for users to change these suffixes for whatever platform they have. Q: will this patch only affect the desired systems, or will scons think .aspp is now assembly even on non-gnu systems? Not sure.
64
Greg1.0.xp2not StevenOfficially 1.0.x, but whoever gets it should apply the patch ASAP.
65
Jim
66
Ken
67
Sohail
68
Steven1.x?not meThere's another issue (very old) re: configurability of these suffixes. Can't find it right now.
69
70
21840DEFECTtriageP4stevenknightGlob pollutes LIBPATH
71
consensus
72
73
Bill
74
Brandon
75
David
76
GaryNot sure I agree w/Greg's comment. Adding the Glob in src/util/SConscript in his testcase does make the LIBPATH behavior *different* (compared to just using ['util.cpp']); it shouldn't. One way or another should be correct, but the Glob shouldn't change it. His patch makes it return a *new* list, not a pointer to the original list (I think?) which seems like it might be on the right track. My recommendation is for someone to look into it for 1.x. The rdirs stuff does get complicated.
77
GregwontfixWell, this has nothing to do with Glob() and I believe the behavior is correct. The expanded LIBPATH should include both the variant dir and the source dir (and repositories, if any) to cover the case where the library is found in the source. Moreover, he shouldn't be putting anything in LIBPATH as he's put the File node in LIBS. And I have no clue what his patch does; I don't see that it should change the behavior. On top of that, if he really only wants -Lbuild/util, he should add Dir('build/util') to LIBPATH.
78
Jim
79
Ken
80
Sohail
81
Steven1.1p3The fix is a good one, he's just making a list copy where it should, instead of a reference to the same list. Without this, you get later uses of the same object changing out from under, just because you call Glob() elsewhere. The behavior should be consistent regardless.
82
83
21850ENHANCEMENTtriageP4bdbaddogAdd Explicit() to scons
84
consensus
85
86
Bill
87
Brandon
88
David
89
Gary1.xmedoc issue. User Manual (new section, Not Building a Subdir By Default) and/or man page (under Ignore). Agree w/ Greg, no separate function.
90
GregwontfixPut it in a wiki page or the User Manual (or both), but I see no need for it as a separate specialized function.
91
Jim
92
Ken
93
Sohail
94
Steven??I really don't have a strong opinion and will go with the flow
95
96
21660DEFECTtriageP2gregnoelCrash using positional varlist in Action()
97
consensus
98
99
Bill
100
Brandon
Loading...
 
 
 
Sheet 1