Issues 2009-08-25
 Share
The version of the browser you are using is no longer supported. Please upgrade to a supported browser.Dismiss

 
View only
 
|
 
Still loading...
ABCDEFG
1
IDVotesType/MemberMilestonePriOwnerSummary/Comments
2
3
24260DEFECTtriageP4issues@sconsCPPScanner incorrectly believes that <> style #include causes compiler search for header in current dir
4
consensus
5
6
BillfutureCan push this to be part of tool redesign?
7
Brandon
8
David
9
Gary3.xShould be configurable, then we could get it right for the 99% of compilers that do the usual thing. Is there a hook in the code to see where the filename came from? If so, we should do it earlier. If not, then it's harder, so defer to 3.x
10
Greginvalid, but 3.x is OKThe distinction between <> and "" is implementation-defined in the C and C++ standards. Even depending on a header to be in a directory is not portable.
11
Jim
12
Ken
13
Sohail
14
Steveninvalidwhat Greg said. although, you can make a case that since the behavior *is* undefined, our enforcing one particular interpretation of <> is potentially at odds with different toolchains that we might want to support, and should therefore be made configurable...?
15
16
24270DEFECTtriageP4issues@sconsSTATIC_AND_SHARED_OBJECTS_ARE_THE_SAME not in man page
17
consensus
18
19
BilldocAgree with Steven
20
Brandon
21
David
22
GarydocEither need this or a supported way to set the "shared" attribute on an obj, and in most cases this hack is easier to use.
23
GreginvalidThis is an internal hack that shouldn't be exposed.

@Steven, I'd have an Ar() or Archive() builder that took any kind of object and could be used for the kinds of things they're trying to do. Don't forget, the original purpose of the `ar` command was to build archives; the use for libraries came later.
24
Jim
25
Ken
26
Sohail
27
Jim
28
Ken
29
Sohail
30
Steven?I don't view it as an internal hack. If we don't document this, then what mechanism would you suggest for controlling this behavior? It's downright necessary to make certain configurations work with Visual C++.
31
32
24280DEFECTtriageP4issues@sconsfails to bootstrap when tar.exe in path
33
consensus 3.x/p4 or future/p?, packaging guru TBD
34
David
35
Bill3.x /futurep4packaging guruditto below (Steven, Gary)
36
Brandon
37
David
38
Gary3.x / futurep4packaging guruMaybe Lukas will be up for this eventually. I suspect it's a simple $PATH issue actually (not even really packaging per se)
39
Greginvalid?Lukas?Who's the expert on packaging? Is it even supposed to work on Cygwin? Would Lukas be a possibility?
40
Jim
41
Ken
42
Sohail
43
Steven3.x / futurep4packaging guruwould be good to make it work under Cygwin, but it isn't exactly high priority
44
45
24290DEFECTtriageP4issues@sconsenvironment overrides lost in call to added method
46
consensus
47
David
48
BillHmm...Are OverrideEnvironments() supported, or just internal?
49
Brandon
50
David
51
GaryDon't know enough about the internals to say, but what can we say to the OP? Wontfix because it's hard? 3.x because we don't have cycles to investigate it now?
52
GregHmmm...He's right, it's unexpected, but I'm not having any flashes of insight as to how to make it work correctly without making the OverrideEnvironment much more expensive.
53
Jim
54
Ken
55
Sohail
56
Stevenresearchwho?probably should fall out of some hypothetical rewrite of environments to clean up substitution and all the other warts
57
58
24300DEFECTtriageP4issues@sconsConfigure() with conf_dir and VariantDir breaks when variant directory deleted.
59
consensus: 2.x p3 SK
60
David
61
Bill2.xp3sk
62
Brandon
63
David
64
Gary2.xp3skthanks Steven
65
Greg2.1?StevenThis is probably another case of crossed assumptions, as when `scons -n` is used with the .sconsign in a not-yet-existing variant dir. Steven solved the last one.
66
Jim
67
Ken
68
Sohail
69
Steven2.xp3sk
70
71
24310PATCHtriageP4issues@sconsAllow customisation of tempfile extension
72
consensus: 2.x p3 +Easy Bill
73
David
74
Bill2.xp3bdeeganI looked at the patch, should it provide defaults in the code if the TEMPFILESUFFIX is not defined, current checks if defined and if not then defaults to hardcoded ".lnk"
75
Brandon
76
David
77
Gary2.xp3 +EasyAgree w/ Steven, easy & useful(Q: wasn't there another recent tempfile-related patch?)
78
Gregno ideadefer to experts
79
Jim
80
Ken
81
Sohail
82
Steven2.xp3 +Easyhaven't looked at the patch, but should be straightforward to add. Should use $TEMPFILESUFFIX, though, not $TEMPFILEEXTENSION
83
84
24320DEFECTtriageP4issues@sconsJava builder sets wrong SRCPATH during package build
85
consensus: 2.x p3 +java TBD
86
David
87
Bill2.x"+java"
88
Brandon +javaReport sounds coherent, but I have no idea.
89
David
90
Gary2.xp3 +javaagree w/ others
91
Greg +javaNeed java expert. He wants to set --srcpath on the command line and then cherry-pick files within the source directory. I don't know if javac can even do that.
92
Jim
93
Ken
94
Sohail
95
Steven2.xp3 +javawhat greg said
96
97
24330PATCHtriageP4issues@sconsAdd dylib_file to ParseFlags
98
consensus: 1.3. p2 Greg
99
100
Bill1.3p2gregHarmless, though a test might be nice.
Loading...
 
 
 
Sheet 1