http://bugzilla.gdcproject.org/show_bug.cgi?id=202

Sebastien Alaiwan <sebastien.alaiwan@gmail.com> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
            Summary|Undefined reference to      |Undefined reference to
                   |libcurl functions with      |libcurl functions with
                   |small program               |small program not
                   |                            |explicitly using network.

-- 
You are receiving this mail because:
You are watching all bug changes.

http://bugzilla.gdcproject.org/show_bug.cgi?id=202

Sebastien Alaiwan <sebastien.alaiwan@gmail.com> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
            Summary|Undefined reference to      |Undefined reference to
                   |libcurl functions with      |libcurl functions with
                   |small program not           |small program (not
                   |explicitly using network.   |explicitly using network).

-- 
You are receiving this mail because:
You are watching all bug changes.

http://bugzilla.gdcproject.org/show_bug.cgi?id=202

--- Comment #1 from Sebastien Alaiwan <sebastien.alaiwan@gmail.com> ---
It seems to occur when std.conv and byLine are used in the same module. Here's another example which causes link issues:

// cantlink.d
int main()
{
  import std.stdio;
  stdin.byLine().front();
  return 0;
}

void f(char[] s)
{
  import std.conv;
  to!int(s);
}

$ gdc cantlink.d

If the 'char[]' is changed to a 'string', the problem disappears.

-- 
You are receiving this mail because:
You are watching all bug changes.

http://bugzilla.gdcproject.org/show_bug.cgi?id=202

--- Comment #2 from Sebastien Alaiwan <sebastien.alaiwan@gmail.com> ---
My version:
gdc (Debian 5.3.1-7) 5.3.1 20160121

-- 
You are receiving this mail because:
You are watching all bug changes.

http://bugzilla.gdcproject.org/show_bug.cgi?id=202

--- Comment #3 from Sebastien Alaiwan <sebastien.alaiwan@gmail.com> ---
I was thinking too long symbol names generated by gdc could cause a memory corruption in the linker.

However, I was able to reproduce the exact same symptoms if I try to link the resulting object file with the "gold" linker (instead of ld).

Valgrind during the linking step shows nothing interesting.

-- 
You are receiving this mail because:
You are watching all bug changes.

http://bugzilla.gdcproject.org/show_bug.cgi?id=202

--- Comment #4 from Iain Buclaw <ibuclaw@gdcproject.org> ---
What version of ld are you using?

I could reproduce the first example when you reported it.  But now I'm struggling to reproduce it at all.

---
GNU ld (GNU Binutils) 2.26.51.20160128
Copyright (C) 2016 Free Software Foundation, Inc.
This program is free software; you may redistribute it under the terms of
the GNU General Public Licence version 3 or (at your option) a later version.
This program has absolutely no warranty.

-- 
You are receiving this mail because:
You are watching all bug changes.

http://bugzilla.gdcproject.org/show_bug.cgi?id=202

--- Comment #5 from Iain Buclaw <ibuclaw@gdcproject.org> ---
It could also be that I'm using the 2.067 branch (PR #99) too.  Maybe something has changed between versions.

-- 
You are receiving this mail because:
You are watching all bug changes.

http://bugzilla.gdcproject.org/show_bug.cgi?id=202

--- Comment #6 from Sebastien Alaiwan <sebastien.alaiwan@gmail.com> ---
Same as you:
$ ld --version
GNU ld (GNU Binutils for Debian) 2.26
Copyright (C) 2015 Free Software Foundation, Inc.
This program is free software; you may redistribute it under the terms of
the GNU General Public License version 3 or (at your option) a later version.
This program has absolutely no warranty.

-- 
You are receiving this mail because:
You are watching all bug changes.

http://bugzilla.gdcproject.org/show_bug.cgi?id=202

--- Comment #7 from Iain Buclaw <ibuclaw@gdcproject.org> ---
Ok. I'll give master a try.

-- 
You are receiving this mail because:
You are watching all bug changes.

http://bugzilla.gdcproject.org/show_bug.cgi?id=202

--- Comment #8 from Sebastien Alaiwan <sebastien.alaiwan@gmail.com> ---
Please let me know if you need some more information / tests from me.

This is starting to become critical on our side ; because although for GNU/Linux targets, it can be worked around by adding '-lcurl' to LDFLAGS (at the expense of making the resulting binary dependent on curl), there seem to be no workaround for mingw target (I get undefined references to Winsock functions, and adding '-lws2_32' doesn't help - maybe a calling convention problem?)

-- 
You are receiving this mail because:
You are watching all bug changes.