Cristian Stoica wrote:
> Hi Fulvio,
>
> Some people take toolchain upgrades less lightly. For example, Gentoo 
> has currently unmasked only gcc 4.6.
Are you trying to argue that it is better to use software that is not 
updated?
There are technical reasons why Gentoo is forced to long release cycles 
for gcc, but they update as soon as they can and even debian stable now 
use gcc 4.7.2

>
> According to http://gcc.gnu.org/projects/cxx0x.html, support for 
> -std=c++0x is less likely to be backported and gcc 4.8 is rather new.
You have misread: I wrote the multi-threading code in 2011 using gcc 
4.6.1 without problems and even your 4.5.4 claim to support it

>
> Since the problem I saw with game window comes from -std=c++0x rather 
> than from -pthreads did you consider activating c++0x only if the 
> toolchain supports it instead of setting it the default?|| What is 
> your view on this?
>
It's not about c++0x support: it is a known bug.
The configure script actually used to activate multi-threading only if 
c++0x was supported (in 2011 gcc versions without c++0x support were not 
unusual), but I have changed the configure script recently precisely 
because of this bug: practically every gcc version around claim to 
support c++0x, but there are old versions that crash at runtime with a 
cryptic message.
So the THREADS option was added and now the decision on when to turn off 
multi-threading is left to the user:
when you run ./configure:
' Multi-threading enabled (if you have problems try ./configure 
THREADS="").  '
in the wiki:
http://sourceforge.net/p/scid/wiki/CompileScid/
and in many posts in the mailing-list.

I think it's a false problem: there are servers not connected to 
internet on which makes sense to not risk updating something that works 
perfectly, or some users who likes his old systems (i also have and 
sometime use an old ubuntu 10.4 with gnome 2) but writing ./configure 
THREADS="" don't seem so complicated to me.

However, recently on the mailing-list, Marius Roets offered to work on a 
better build process: you can work with him to write a portable way 
(should work on Macs too with clang) to detect the bug and automatically 
disable multi-threading (just avoid some ugly if gcc -v < 4.6 then).

Bye,
Fulvio

------------------------------------------------------------------------------
Try New Relic Now & We'll Send You this Cool Shirt
New Relic is the only SaaS-based application performance monitoring service 
that delivers powerful full stack analytics. Optimize and monitor your
browser, app, & servers with just a few lines of code. Try New Relic
and get this awesome Nerd Life shirt! http://p.sf.net/sfu/newrelic_d2d_may
_______________________________________________
Scid-users mailing list
Scid-users@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/scid-users

Reply via email to