Hi, > >>> I've made a local change to the 1.2 sources to add a 'batch' attribute to > >>> the execon task (telling it how many source files to pass per batch > >>> execution of the command). Should I submit this patch? > >> > >> Sounds all right to me. Some of the underlying executable programs may > >> also expect limited number of arguments, so your patch would make that > >> easier to use as well. > > > > > > I think it would be preferable to specify the maximum length of the > > command rather than the number of arguments. Since length of paths can > > vary wildly, there's no guarantee a certain number of arguments won't > > still be too long in some cases. > True. In fact that was my first thought, but the number of arguments was > a lot easier to implement and I was sick of 1/2 hour builds with > parallel=false 8^}). (my haven't I become impatient!) > > Although limiting only the length of the command wouldn't help with the > situation Mariusz describes, with an external command only capable of > groking a particular number...
Limiting a length is such an artificial and OS-dependent-only thing, that I would rather not make it a parameter of a build target, but place it (i.e. hardcode it) inside the execon implementation for a particular OS. If for example given OS cannot cope with command lines longer than given limit then hardcode it into the target that if OS==givenOS and length>givenLimit, split it. What do you think? cheers Mariusz
