www.digitalmars.com         C & C++   DMDScript  

digitalmars.D - Sooo, umm, status of Optlink Issue #6144?

reply Nick Sabalausky <SeeWebsiteToContactMe semitwist.com> writes:
Regarding the blocker "Unexpected OPTLINK Termination at EIP=00428DA3":
http://d.puremagic.com/issues/show_bug.cgi?id=6144

I'm hitting on that issue now too, and it's a royal PITA since I can't
even get stack traces, there's no obvious way to de-trigger the
problem and the problem is known to resist minimized test cases.

So what's the current status of it (which I ask hoping the answer is
something like "a top priority being actively worked on")?

More importantly, is there any info/files/etc I can provide to help
out, or is everything needed already known and merely awaiting a fix
from anyone who knows much more about optlink than I do?
Apr 08 2013
next sibling parent reply Walter Bright <newshound2 digitalmars.com> writes:
On 4/8/2013 12:33 PM, Nick Sabalausky wrote:
 Regarding the blocker "Unexpected OPTLINK Termination at EIP=00428DA3":
 http://d.puremagic.com/issues/show_bug.cgi?id=6144

 I'm hitting on that issue now too, and it's a royal PITA since I can't
 even get stack traces, there's no obvious way to de-trigger the
 problem and the problem is known to resist minimized test cases.

 So what's the current status of it (which I ask hoping the answer is
 something like "a top priority being actively worked on")?

 More importantly, is there any info/files/etc I can provide to help
 out, or is everything needed already known and merely awaiting a fix
 from anyone who knows much more about optlink than I do?

Denis Shelomovskij has a pull request for optlink on it: https://github.com/DigitalMars/optlink/pull/8 You can give that a try. My main difficulty with it is it's a hack. But if it works for you, that might be good enough.
Apr 08 2013
parent reply Denis Shelomovskij <verylonglogin.reg gmail.com> writes:
09.04.2013 1:16, Walter Bright пишет:
 On 4/8/2013 12:33 PM, Nick Sabalausky wrote:
 Regarding the blocker "Unexpected OPTLINK Termination at EIP=00428DA3":
 http://d.puremagic.com/issues/show_bug.cgi?id=6144

 I'm hitting on that issue now too, and it's a royal PITA since I can't
 even get stack traces, there's no obvious way to de-trigger the
 problem and the problem is known to resist minimized test cases.

 So what's the current status of it (which I ask hoping the answer is
 something like "a top priority being actively worked on")?

 More importantly, is there any info/files/etc I can provide to help
 out, or is everything needed already known and merely awaiting a fix
 from anyone who knows much more about optlink than I do?

Denis Shelomovskij has a pull request for optlink on it: https://github.com/DigitalMars/optlink/pull/8 You can give that a try. My main difficulty with it is it's a hack. But if it works for you, that might be good enough.

It works for me. If it will work for Nick too will you merge pull #8, resolving 6144 and open a new ticket about underlying issue? -- Денис В. Шеломовский Denis V. Shelomovskij
Apr 09 2013
parent Andrei Alexandrescu <SeeWebsiteForEmail erdani.org> writes:
On 4/9/13 6:53 AM, Denis Shelomovskij wrote:
 09.04.2013 1:16, Walter Bright пишет:
 On 4/8/2013 12:33 PM, Nick Sabalausky wrote:
 Regarding the blocker "Unexpected OPTLINK Termination at EIP=00428DA3":
 http://d.puremagic.com/issues/show_bug.cgi?id=6144

 I'm hitting on that issue now too, and it's a royal PITA since I can't
 even get stack traces, there's no obvious way to de-trigger the
 problem and the problem is known to resist minimized test cases.

 So what's the current status of it (which I ask hoping the answer is
 something like "a top priority being actively worked on")?

 More importantly, is there any info/files/etc I can provide to help
 out, or is everything needed already known and merely awaiting a fix
 from anyone who knows much more about optlink than I do?

Denis Shelomovskij has a pull request for optlink on it: https://github.com/DigitalMars/optlink/pull/8 You can give that a try. My main difficulty with it is it's a hack. But if it works for you, that might be good enough.

It works for me. If it will work for Nick too will you merge pull #8, resolving 6144 and open a new ticket about underlying issue?

I think that's a good step. And let's not forget that bugs etc. can always be reopened. Andrei
Apr 09 2013
prev sibling next sibling parent reply Denis Shelomovskij <verylonglogin.reg gmail.com> writes:
08.04.2013 23:33, Nick Sabalausky пишет:
 Regarding the blocker "Unexpected OPTLINK Termination at EIP=00428DA3":
 http://d.puremagic.com/issues/show_bug.cgi?id=6144

 I'm hitting on that issue now too, and it's a royal PITA since I can't
 even get stack traces, there's no obvious way to de-trigger the
 problem and the problem is known to resist minimized test cases.

 So what's the current status of it (which I ask hoping the answer is
 something like "a top priority being actively worked on")?

 More importantly, is there any info/files/etc I can provide to help
 out, or is everything needed already known and merely awaiting a fix
 from anyone who knows much more about optlink than I do?

If you are not interesting in developing OPTLINK but want to live without Issue 6144, email me (on this address or without ".reg" suffix) and I'll send you fixed OPTLINK. If you want to be able to build OPTLINK (I mean without dmc) you need pull #4, to debug and use Visual C++ you need pull #6 (both are ignored ~20 days so I don't expect them to be merged soon). And #8 to fix Issue 6144, but not probably existed underlying OPTLINK bug which I hope will not cause problems in near future (looks like Walter is against such solution so don't expect it to be merged [soon]). To make things easier use "patched" branch from my fork with everything merged: https://github.com/denis-sh/optlink/tree/patched Do not forget: you also need `imagecfg` [1] and prepend `build_optlink.bat` with something like this: --- set DM=D:\D\dm set PATH=%DM%\bin;%PATH% --- [1] http://www.robpol86.com/index.php/ImageCFG -- Денис В. Шеломовский Denis V. Shelomovskij
Apr 09 2013
next sibling parent "Rene Zwanenburg" <renezwanenburg gmail.com> writes:
On Tuesday, 9 April 2013 at 11:10:40 UTC, Denis Shelomovskij 
wrote:
 If you are not interesting in developing OPTLINK but want to 
 live without Issue 6144, email me (on this address or without 
 ".reg" suffix) and I'll send you fixed OPTLINK.

Hi, It looks like the web interface to the NG doesn't show email addresses (good thing), so I'm posting here instead of mailing you. I'd greatly appreciate it if you could send me the fixed binary. The address I'm using for this post is valid.
Apr 09 2013
prev sibling parent reply Walter Bright <newshound2 digitalmars.com> writes:
On 4/9/2013 4:10 AM, Denis Shelomovskij wrote:
 If you are not interesting in developing OPTLINK but want to live without Issue
 6144, email me (on this address or without ".reg" suffix) and I'll send you
 fixed OPTLINK.

Denis, I have neglected to thank you for your help with this, and your willingness to help others. Thanks! I really do appreciate it.
Apr 09 2013
parent reply Denis Shelomovskij <verylonglogin.reg gmail.com> writes:
10.04.2013 0:44, Walter Bright пишет:
 On 4/9/2013 4:10 AM, Denis Shelomovskij wrote:
 If you are not interesting in developing OPTLINK but want to live
 without Issue
 6144, email me (on this address or without ".reg" suffix) and I'll
 send you
 fixed OPTLINK.

Denis, I have neglected to thank you for your help with this, and your willingness to help others. Thanks! I really do appreciate it.

First of all, thanks for the thanks. But I have to say I see the only one reason for it here. It is "Wow, that nasty person who used to insult everybody just did something useful. Lets say "Thanks" hoping he will like it and begin being useful". So I'd like, if you don't mind, to redirect the "Thanks" to the (really time wasting and useful in a context of recent "shared library" talks and changes) fact I fixed TLS loading on Windows XP (yes, just finished what Rainer started) thus allowing languages like D to load/unload DLL-s dynamically. -- Денис В. Шеломовский Denis V. Shelomovskij
Apr 09 2013
parent reply Walter Bright <newshound2 digitalmars.com> writes:
On 4/9/2013 2:41 PM, Denis Shelomovskij wrote:
 But I have to say I see the only one reason for it here. It is "Wow, that nasty
 person who used to insult everybody just did something useful. Lets say
"Thanks"
 hoping he will like it and begin being useful".

You've been doing valuable stuff all along. It's why we put up with you :-) I would, however, prefer that when you fire a broadside that you direct it at me rather than others here.
 So I'd like, if you don't mind, to redirect the "Thanks" to the (really time
 wasting and useful in a context of recent "shared library" talks and changes)
 fact I fixed TLS loading on Windows XP (yes, just finished what Rainer started)
 thus allowing languages like D to load/unload DLL-s dynamically.

This is great news. Can you point me to the pull req on that?
Apr 09 2013
parent Denis Shelomovskij <verylonglogin.reg gmail.com> writes:
10.04.2013 2:20, Walter Bright пишет:
 On 4/9/2013 2:41 PM, Denis Shelomovskij wrote:
 But I have to say I see the only one reason for it here. It is "Wow,
 that nasty
 person who used to insult everybody just did something useful. Lets
 say "Thanks"
 hoping he will like it and begin being useful".

You've been doing valuable stuff all along. It's why we put up with you :-) I would, however, prefer that when you fire a broadside that you direct it at me rather than others here.
 So I'd like, if you don't mind, to redirect the "Thanks" to the
 (really time
 wasting and useful in a context of recent "shared library" talks and
 changes)
 fact I fixed TLS loading on Windows XP (yes, just finished what Rainer
 started)
 thus allowing languages like D to load/unload DLL-s dynamically.

This is great news. Can you point me to the pull req on that?

No pulls, it isn't a part of druntime. It can be but only in the case executable is compiled. For D DLL it have to be loaded before the DLL and must not be unloaded until process termination. NG post: http://forum.dlang.org/thread/kf877d$iqd$1 digitalmars.com NG post notes: * replace `snn.dll` with `snn.lib` * according the injector, regressions are easily fought but Issue 9044 prevented it from being able to build -- Денис В. Шеломовский Denis V. Shelomovskij
Apr 10 2013
prev sibling next sibling parent Nick Sabalausky <SeeWebsiteToContactMe semitwist.com> writes:
On Tue, 09 Apr 2013 14:53:57 +0400
Denis Shelomovskij <verylonglogin.reg gmail.com> wrote:

 09.04.2013 1:16, Walter Bright =D0=BF=D0=B8=D1=88=D0=B5=D1=82:
 Denis Shelomovskij has a pull request for optlink on it:

 https://github.com/DigitalMars/optlink/pull/8

 You can give that a try. My main difficulty with it is it's a hack.
 But if it works for you, that might be good enough.

It works for me. If it will work for Nick too will you merge pull #8,=20 resolving 6144 and open a new ticket about underlying issue? =20

Works for me, too. :)
Apr 09 2013
prev sibling parent reply Denis Shelomovskij <verylonglogin.reg gmail.com> writes:
08.04.2013 23:33, Nick Sabalausky пишет:
 Regarding the blocker "Unexpected OPTLINK Termination at EIP=00428DA3":
 http://d.puremagic.com/issues/show_bug.cgi?id=6144

 I'm hitting on that issue now too, and it's a royal PITA since I can't
 even get stack traces, there's no obvious way to de-trigger the
 problem and the problem is known to resist minimized test cases.

 So what's the current status of it (which I ask hoping the answer is
 something like "a top priority being actively worked on")?

 More importantly, is there any info/files/etc I can provide to help
 out, or is everything needed already known and merely awaiting a fix
 from anyone who knows much more about optlink than I do?

Walter Bright properly fixed the issue. See http://d.puremagic.com/issues/show_bug.cgi?id=6144#c15 for the fix and fixed OPTLINK build. -- Денис В. Шеломовский Denis V. Shelomovskij
Apr 10 2013
parent "thedeemon" <dlang thedeemon.com> writes:
On Wednesday, 10 April 2013 at 10:27:07 UTC, Denis Shelomovskij 
wrote:

 Walter Bright properly fixed the issue. See 
 http://d.puremagic.com/issues/show_bug.cgi?id=6144#c15 for the 
 fix and fixed OPTLINK build.

Hooray! I had the same problem in a recent project [1] and can confirm now the linker doesn't crash anymore. [1]: https://bitbucket.org/infognition/undup
Apr 10 2013