Message-ID: <8D53104ECD0CD211AF4000A0C9D60AE33CA3A2@probe-2.Acclaim-Euro.net> From: Shawn Hargreaves To: djgpp AT delorie DOT com Cc: Nate Eldredge Subject: Re: Announce: Allegro 3.1 Date: Wed, 6 Jan 1999 15:44:08 -0000 MIME-Version: 1.0 X-Mailer: Internet Mail Service (5.0.1460.8) Content-Type: text/plain Reply-To: djgpp AT delorie DOT com Nate Eldredge writes: > It makes sense to me that anyone who wants to support more than one > version should stay away from -Werror in the distribution Makefile. You are probably right about that, but it goes against the grain for me to just ignore these warnings. I want fairly strict code scrutiny, because I think that most compiler warnings do give useful information, but there comes a point where the warnings have become too pedantic and stupid, and I am no longer interested in fixing their problems. Obviously exactly where this cutoff point is located will differ from one person to another, which is why it would be cool if there was some reliable way to enable only those warnings that you are specifically concerned about. I thought I'd found the settings that did this for me, so it is disappointing to find that egcs interprets these switches differently to gcc. I use -Werror in my code because I want to know immediately if any warnings do crop up, and I left it in the release makefile partly because it didn't occur to me that this could ever be a problem, but also for the sake of other Allegro developers (most will probably want an instant response to any warnings, and for those who usually just ignore warning output, this is a useful hint that I want all the code to build cleanly :-) Since this now seems to be a problem, though, I agree that it should be changed to only use -Werror if that has specifically been enabled. > And how is a compiler to recognize an option to disable a warning which > won't exist until a future version? Obviously it cannot :-) That is why I think these new warnings should not be enabled by default: it is easy for users of that specific compiler to enable them by using an extra switch, but impossible for them to be turned off by the authors of code that must support multiple compiler versions... Shawn Hargreaves.