Bug 11321 – Can't link _D6object15__T7reserveTyaZ7reserveFNaNbNeKAyamZm

Status
RESOLVED
Resolution
FIXED
Severity
regression
Priority
P2
Component
dmd
Product
D
Version
D2
Platform
x86_64
OS
Linux
Creation time
2013-10-21T23:20:00Z
Last change time
2014-02-13T13:32:19Z
Keywords
bounty
Assigned to
nobody
Creator
deadalnix

Attachments

IDFilenameSummaryContent-TypeSize
1275linkfail.zipUnzip, compile and link fail.application/x-zip8032

Comments

Comment #0 by deadalnix — 2013-10-21T23:20:35Z
Created attachment 1275 Unzip, compile and link fail. See zip file attached. make DMD=PATH_TO_DMD to compile. It does link fail. The error do not show up when not using separate compilation. Sorry to not be able to reduce more, I really short of time right now, but I guess it is better if the bug report come before the release.
Comment #1 by bugzilla — 2013-10-26T13:28:13Z
Can you try it with -allinst ?
Comment #2 by bugzilla — 2013-10-27T14:56:51Z
This works with -allinst flag.
Comment #3 by public — 2013-10-27T15:48:58Z
Exactly the attitude I feared. "I won't figure out why this does not work with new feature because legacy behavior workaround works".
Comment #4 by bugzilla — 2013-10-27T21:27:35Z
There's only one of me. I welcome any help in reducing this example to something more tractable.
Comment #5 by deadalnix — 2013-10-27T21:57:47Z
(In reply to comment #2) > This works with -allinst flag. I'm sorry, I simply can't reduce that further now. I really do not have time. I'll have time to do it next week. It seems that the safe thing to do here is to go for allinst as default behavior and introduce the new behavior as a flag. We can switch to us this new behavior by default with the next release.
Comment #6 by braddr — 2013-10-27T22:50:44Z
Reopening. Though, we may choose to ship with this regression since it has something of a work around.
Comment #7 by public — 2013-10-28T04:47:40Z
(In reply to comment #4) > There's only one of me. I welcome any help in reducing this example to > something more tractable. Sure, but is not a legitimate reason to close a valid (though incomplete) bug report with RESOLVED/WORKSFORME.
Comment #8 by Daniel.Wyatt — 2013-11-22T20:48:34Z
This seems to be fixed in 46b294ff2d (fixes 11239). Someone confirm & close?
Comment #9 by dmitry.olsh — 2014-01-07T07:13:28Z
Works for me after fixing Location --> Location() in semantic.d. Also I used dmd as the driver for linking to avoid messing with makefile paths.
Comment #10 by deadalnix — 2014-01-10T22:15:00Z
(In reply to comment #8) > This seems to be fixed in 46b294ff2d (fixes 11239). > Someone confirm & close? I suspect this has changed the behavior of the bug, but didn't solved it. I'll run a Dustmite over the weekend and see where it goes.
Comment #11 by deadalnix — 2014-02-13T13:32:19Z
(In reply to comment #10) > (In reply to comment #8) > > This seems to be fixed in 46b294ff2d (fixes 11239). > > Someone confirm & close? > > I suspect this has changed the behavior of the bug, but didn't solved it. I'll > run a Dustmite over the weekend and see where it goes. Forgot to update the bug report. The bug do not manifest in the beta released recently. This can be closed.