Re: guavac bug #22325
On Wed, Jun 10, 1998 at 10:05:05AM -0400, Brian White wrote: Bug #22325, marked important, says that my package guavac has an unsatisfied suggestion on java-virtual-machine. I need your advice on what to do about it. I'd do either: 1) ignore the suggests problem I think this will have to be the plan of action. It doesn't build in g++ 2.8 with libstdc++2.8/libg++2.8, it almost builds on g++272 with libg++272-dev but not quite, presumably because the autoconf stuff tests some things using `g++' (2.8) but then the code gets compiled with g++272, and I'm not too sure how to work around this. Unfortunately we don't really seem to support concurrent 2.7.2 and 2.8 development, since the -dev packages conflict. 2) repack the existing package to remove the suggests line. How do you mean `repack'? Since it's a valid virtual package name I'm not even convinced it's a bug in any package at all. May I change it to wishlist? thanks, Hamish -- Hamish Moffatt, [EMAIL PROTECTED], [EMAIL PROTECTED], [EMAIL PROTECTED] Latest Debian packages at ftp://ftp.rising.com.au/pub/hamish. PGP#EFA6B9D5 CCs of replies from mailing lists are welcome. http://hamish.home.ml.org -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of unsubscribe. Trouble? Contact [EMAIL PROTECTED]
Re: guavac bug #22325
Bug #22325, marked important, says that my package guavac has an unsatisfied suggestion on java-virtual-machine. I need your advice on what to do about it. My thoughts are: 1. It's a suggestion only, so nothing will break if it doesn't exist. Unfortunately dselect is a bit picky about even suggestions. At the time I last built it, kaffe existed [albeit it in contrib] which satisfied the suggestion. (Interestingly there's my reported bug #16652 IIRC which says that javalex should move to contrib since it depends: java-virtual-machine -- but even that move won't solve its deps now). Perhaps there's the standard bug in dselect, `dselect is not user friendly', here. 2. It's a bug in some other package that java-virtual-machine doesn't exist. It's a registered virtual package name, AFAIK, so at one stage it must have been provided. jdk1.1-runtime should probably provide it but doesn't [bug?]. 3. I could built it again without the suggestion. However the suggestion is valid, and more to the point, with the C++ compiler changeover it doesn't build out of the box. This is bad in its own right, but I think not `important'. I will have to spend some time tweaking it, and then hope that it still works as a result. I would prefer to avoid this. I'd do either: 1) ignore the suggests problem 2) repack the existing package to remove the suggests line. Brian ( [EMAIL PROTECTED] ) --- Friends are relatives that you make for yourself. -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of unsubscribe. Trouble? Contact [EMAIL PROTECTED]