Re: [Boost-users] thread no defined - issue

Subject: Re: [Boost-users] thread no defined - issue
From: Gavin Lambert (gavinl_at_[hidden])
Date: 2016-09-14 23:34:04

On 10/09/2016 20:39, "Alexander Carôt" wrote:
>> How have you built boost and how your user has build boost?
>> Which compile time flags are you using in your application and which is
>> using your user?
>> Are you defining BOOST_THREAD_POSIX
>
> No, not at all: I am neither compiling boost with additional flags
> noram I defining anything in my sources.
>
> The user simply runs a binary, which tells that threads are not
> available and hence does not work at all.

The code that you highlighted is based on a preprocessor symbol, so it 
is impossible for this to trigger on one machine and not another unless 
they are recompiling the sources rather than using a pre-built binary.

> So - maybe this is where the problem starts: Should I explicitely
> compile boost with thread-support (or even pthread-support) ?

Boost should be compiled with thread support by default.  This thread 
support might be native Windows rather than pthreads.  Ideally, you 
should not care which of these are used, and should stick to the common 
Boost.Thread APIs that work on both.

So probably the code that you highlighted is checking the wrong thing, 
since it's looking for something pthreads-specific.

As I said before, the best thing to do is to change your code so that 
it's not pthreads-specific by removing use of native/pthreads APIs -- or 
failing that (if there's some reason you absolutely must use the native 
API for something), use a better method to identify whether pthreads or 
Windows threads are used and do the appropriate native thing in each case.

Always forcing use of pthreads is another option, but it's a less-good 
option.  You might be able to get away with it for an application, but 
it's a very bad idea if you're making a library or something the user is 
expected to compile themselves.