Re: Please help freeing libcolt-java
Thorsten Glaser wrote:
Can you, or anyone really, give some detail on what exactly is the
problem here (which files)?
I'm sorry for not having been more specific from the beginning. Given
that you had brought up the interface aspect yourself, I thought we were
on the same page.
Anyway, here my specific concern: Some types (be it implementation
classes directly or Java interfaces) of hep.aida.* appear (as parameter
types and/or return types) in the signatures of public functions of
libcolt. This occurs in the following places:
* toTitleString (696): hep.aida.bin.BinFunction1D
* toTitleString (774): hep.aida.bin.BinFunction1D
* sort (432): hep.aida.bin.BinFunction1D
* aggregate (148): hep.aida.bin.BinFunction1D
* bin (215): hep.aida.bin.DynamicBin1D
* cube (320): hep.aida.IHistogram2D
* cube (359): hep.aida.IHistogram3D
* histogram (497): hep.aida.IHistogram1D
* histogram (508): hep.aida.IHistogram2D
* histogram (520): hep.aida.IHistogram2D
* histogram (532): hep.aida.IHistogram3D
This makes it impossible to separate the interfaces (meaning the set of
signatures that the library provides) of libcolt from the one of aida.
Either you keep the full signatures (the "interface") of
hep.aida.IHistogram1/2/3D, hep.aida.bin.DynamicBin1D and
hep.aida.bin.BinFunction1D with changed implementations - which might be
a legal problem, that's what I'm not sure of. Or we use alternative
parameter/return types - but that breaks the interface of libcolt:
Reverse dependencies might have to be patched in order to work with the
method signatures provided by the new types.
Do not bother with debian-legal on this. They are no body of Debian,
just a mailing list with parties interested in discussing legal
issues, not even lawyers, and often disagreeing with the formal
Good to know, thanks.