|
From: Chris B. <ch...@cn...> - 2007-11-03 17:08:03
|
Yes, I agree with if SoX can not completely find all needed items then it should disable it. If you can, please write a bug report on our bug tracker for each item that find that doesn't work as you described. Then we can fix it as time permits. BTW, the following is my expected behavior for this. It must find usable libraries and headers in normal case or disable the feature. If user specifies --enable-blah and it can not find both libraries and headers then configure should stop executing at that point instead of of just disabling the feature. I don't think we work like that consistently right now but its the ideal way for distribution packages so that they can know when the package they built doesn't have the feature they force-enabled. Chris Michael Chapman wrote: > Just a comment. > > If one has 'something' the "./configure" detects it. > One then runs "make" and it fails if you do not have the development > library/header files. > > E.g. for 'something' = mad/lame/ > > So one can run " ./configure --without-mad --without-lame" > (or install the missing files). > > No problem. > > But a couple of years ago I would not have understood the output of make and > would have just given up. A pity as SoX is wonderful. > > For a more naïve user base, it would be good if configure not only checked > whether lame, mad, etc. were present but also whether they were usable. (Also > configure runs quicker than (as well as before) make on my ancient machine > ;-); > > Just a thought ... if, anyway, it is a practicable thought? > > Michael Chapman > > > |