W liście z pią, 06-02-2004, godz. 05:29, Daniel Silverstone pisze: > On Fri, 2004-02-06 at 03:27, Adam Majer wrote: > > Why do we need this in Debian? It seems from the description that > > sablevm-test-suite is something that is specific to SableVM and > > is used by the SableVM developers to prevent regression bugs, etc.. > > This is, indeed, the nub of the problem and the point of my enquiry. I understand. > > Could you explain why should this be in the archive? Who would > > use this package? > > This is what I wanted to know. I guess my REJECT message wasn't clear > enough. > I get the impression Grzegorz really doesn't want it integrated into the > sablevm package and if the package really is of use to people other than > JVM developers then okay. But if it *really* is just a jvm tester with > no use to end-users, then it doesn't make sense to put it into Debian. Ok. I don't want to pretend that sablevm-test-suite is currently something more useful to the users than gcc tests for gcc users or libffi tests for libffi users, which "only" assure that gcc problems and libffi problems will be hopefuly catched before they hit the users. The difference here is that putting this testing suite in sablevm package is pretty much impossible to do sanely. Cite from my other mail: - if I wanted the test suite to be in SableVM package then it has to be compiled during the build (and for each build separately) as I cannot include binary .classes, - compilation requires not only javac (on which SableVM currently doesn't build-depend, but let's say it wouldn't be that big problem), but also jasmin-sable (it's jasmin java assembler maintained by sable group, as the original author is not interested in it anymore) - jasmin is packaged as a .deb and is written in java, so it Depends: on a JVM. So it would mean that to compile SableVM JVM package - I would already need to have a running JVM, which is kinda sick. Or I would need to include jasmin source into SableVM source and compile jasmin (which compilation btw. also requires working JVM), which I percieve as an insane idea. The thing is that I am simply unable to manually test before each upload whether SableVM works well on all architectures. I spent last 5 months (= ~3 fulltime months) on porting SableVM to these architectures, ensuring that it works properly and it's still not 100% finished. Therefore *I HAVE TO TEST IT* on each architecture, to know whether there are problems, what they are and to work on removing them. I hope you have no doubts about that. Packaging Java is enough PITA already, compared ex. to C/C++, and I simply find no sane way to solve the problems I mentioned above, w/o creating this package. So I need a *binary* package anyway. If this package is not let into archive as a *source* package - I'll have to make it part of some other source, and make a distinct *binary* package out of it anyway. So the amount of software source, amount of binary packages is the same. It's just a question whether you allow me to sanely put distinct things into distinct *source* packages or not. Please try to understand the above technical reasoning behind my request. HTH Grzegorz B. Prokopski -- Grzegorz B. Prokopski <gadek@debian.org> Debian GNU/Linux http://www.debian.org SableVM - LGPLed JVM http://www.sablevm.org Why SableVM ?!? http://devel.sablevm.org/wiki/WhySableVM
Attachment:
signature.asc
Description: To jest =?iso-8859-2?Q?cz=EA=B6=E6?= listu podpisana cyfrowo