On Mon, Oct 26, 2009 at 01:17:12AM +0100, Norbert Preining wrote: > Hi, > > On So, 25 Okt 2009, Guillem Jover wrote: > > There's some packages which rely on HOME existing and often being > > writtable. This is broken behaviour as the package does not have any > > businesss verifying if it exists or writting outside of its build dir > > (or /tmp). > > I disagree. Some upstream build mechanism simply want to do some > checks. Do we want to force each maintainer to make sure this doesn't > happen and patch it away?
Yes. No build system has any business poking around in $HOME. For any reason. If I'm building a package, the contents of my $HOME are irrelevant. Why should my *personal* configuration and other stuff influence *system*- and (for upload) *distribution*-wide packaging? Clearly for consistency and safety, $HOME use should be forbidden entirely. > I have been bitten by that with jppy which runs test suites at built time > using scons, and needs to initialize some data there for test building. That's clearly broken, any data created during the build (for whatever purpose) should be within the build tree. What about consistency between builds, idempotency and other programs altering stuff during builds? > I would suggest on the contrary that HOME *will* be set by all scripts > to a newly created empty directory. While this would alleviate the problem, it's still just papering over the brokenness. IMO we should fail hard in these cases and get the broken build systems fixed. Regards, Roger -- .''`. Roger Leigh : :' : Debian GNU/Linux http://people.debian.org/~rleigh/ `. `' Printing on GNU/Linux? http://gutenprint.sourceforge.net/ `- GPG Public Key: 0x25BFB848 Please GPG sign your mail.
signature.asc
Description: Digital signature