Re: [Scons-dev] Clang support

2015-01-05 Thread Bill Deegan
Looks like your clang tool is not properly detecting that the tool is not there and indicating that.. On Mon, Jan 5, 2015 at 5:56 PM, Paweł Tomulik wrote: > Thanks Bill, > > I took a look at manpage and user docs. It says how we may "specify tools" > and not much more. > > As I understand, this

Re: [Scons-dev] Clang support

2015-01-05 Thread Gary Oberbrunner
On Mon, Jan 5, 2015 at 7:02 PM, Paweł Tomulik wrote: > Looks like SCons is missing a "tool preference system", where each user > (developer, not end user) could easily re-define by its own the preferred > order of compiler toolchains. The same applies to other tools. Don't worry, > there will alw

Re: [Scons-dev] Clang support

2015-01-05 Thread Paweł Tomulik
Thanks Bill, I took a look at manpage and user docs. It says how we may "specify tools" and not much more. As I understand, this allows us to specify what tools are **required** for our project (sorry, this is not clearly stated by docs so I may be wrong here). This doesn't seem to provide

Re: [Scons-dev] Clang support

2015-01-05 Thread Bill Deegan
Pawel, Take a look at Environment() in the Man Page and look at it's tools argument. There should also be info in the users guide. The default behavior when no tools= is specified is supposed to be reasonable, but developers can always do what they want. Often I use Environment(tools=[]) and then

Re: [Scons-dev] Clang support

2015-01-05 Thread Paweł Tomulik
Well, I thought that there were just defaults for given platform, for example gcc for Linux (so it used gcc, if it's installed or fails if it's not, even if other supported compilers are available)?. Am I wrong? W dniu 06.01.2015 o 01:08, Bill Deegan pisze: Pawel, It's always been possible to

Re: [Scons-dev] Clang support

2015-01-05 Thread Bill Deegan
Pawel, It's always been possible to set tool preference in the Environment() creation. As far as allowing a user to override such via local settings, that's be up to the project using SCons. Allowing such by default would likely cause more issues than it solves as one of the core functional requi

Re: [Scons-dev] Clang support

2015-01-05 Thread Michael Jarvis
Clang can (mostly) emulate gcc, while the reverse is not true. I would say default to gcc as well. On Mon, Jan 5, 2015 at 6:00 PM, Bill Deegan wrote: > I'd say on linux default to gcc. > If we add clang tools the user can always override them if they wish to. > > -Bill > > On Mon, Jan 5, 2015 at

Re: [Scons-dev] Clang support

2015-01-05 Thread Paweł Tomulik
Looks like SCons is missing a "tool preference system", where each user (developer, not end user) could easily re-define by its own the preferred order of compiler toolchains. The same applies to other tools. Don't worry, there will always be room for discussion, for example "what should be the

Re: [Scons-dev] Clang support

2015-01-05 Thread Bill Deegan
I'd say on linux default to gcc. If we add clang tools the user can always override them if they wish to. -Bill On Mon, Jan 5, 2015 at 3:58 PM, William Blevins wrote: > Im not sure what percentage of linux devs use clang vs gcc, but my > personal experience is gcc is more widely used. > > Yet a

Re: [Scons-dev] Clang support

2015-01-05 Thread Michael Jarvis
I use both clang and gcc. They both have their pros and cons. On Mon, Jan 5, 2015 at 5:58 PM, William Blevins wrote: > Im not sure what percentage of linux devs use clang vs gcc, but my > personal experience is gcc is more widely used. > > Yet another gcc user, > William > On Jan 5, 2015 6:51

Re: [Scons-dev] Clang support

2015-01-05 Thread William Blevins
Im not sure what percentage of linux devs use clang vs gcc, but my personal experience is gcc is more widely used. Yet another gcc user, William On Jan 5, 2015 6:51 PM, "Russel Winder" wrote: > On Mon, 2015-01-05 at 13:48 +0100, Paweł Tomulik wrote: > […] > > I have a project where I just set co

Re: [Scons-dev] Clang support

2015-01-05 Thread Russel Winder
On Mon, 2015-01-05 at 13:48 +0100, Paweł Tomulik wrote: […] > I have a project where I just set construction variables CC=clang and > CXX=clang++ and it works well (I check existence of these compilers with > SConf, so I don't need the Tool machinery to search for the compiler > executables). >

Re: [Scons-dev] Clang support

2015-01-05 Thread Paweł Tomulik
W dniu 05.01.2015 o 18:09, Dirk Bächle pisze: Hi Paweł, On 05.01.2015 13:48, Paweł Tomulik wrote: W dniu 05.01.2015 o 12:48, Russel Winder pisze: Is it my imagination or is there no support for Clang in SCons? ___ Scons-dev mailing list Scons-dev@

Re: [Scons-dev] Clang support

2015-01-05 Thread Dirk Bächle
Hi Paweł, On 05.01.2015 13:48, Paweł Tomulik wrote: W dniu 05.01.2015 o 12:48, Russel Winder pisze: Is it my imagination or is there no support for Clang in SCons? ___ Scons-dev mailing list Scons-dev@scons.org https://pairlist2.pair.net/mailman/li

Re: [Scons-dev] Clang support

2015-01-05 Thread Gary Oberbrunner
On Mon, Jan 5, 2015 at 7:48 AM, Paweł Tomulik wrote: > ... > I have a project where I just set construction variables CC=clang and > CXX=clang++ and it works well > That's more or less what we do too (in addition to some clang-specific flags we need). Seems to work fine. It'll be much easier

Re: [Scons-dev] Clang support

2015-01-05 Thread Paweł Tomulik
W dniu 05.01.2015 o 12:48, Russel Winder pisze: Is it my imagination or is there no support for Clang in SCons? ___ Scons-dev mailing list Scons-dev@scons.org https://pairlist2.pair.net/mailman/listinfo/scons-dev I have a project where I just set c

Re: [Scons-dev] Clang support

2015-01-05 Thread Damian Maguire
There's no official support that I've ever seen for it, but because Clang generally plays nice with GCC it's usually possible to make it work just by tweaking the environment a little. I actually wrote some small tools as part of a project I work on (OpenMAMA) to support clang, and the static analy

[Scons-dev] Clang support

2015-01-05 Thread Russel Winder
Is it my imagination or is there no support for Clang in SCons? -- Russel. = Dr Russel Winder t: +44 20 7585 2200 voip: sip:russel.win...@ekiga.net 41 Buckmaster Roadm: +44 7770 465 077 xmpp: rus...@winder.or