Re: [Hamlib-developer] libltdl / FT747GX timeout problem - solved?
Library to control radio transceivers and receivers
Brought to you by:
n0nb
|
From: Stephane F. <f8...@fr...> - 2010-04-04 21:06:36
|
Kamal Mostafa skribis: > Stephane Fillod wrote: > > Looks like there's a problem with the embedded libltdl. > > [...] can you install libltdl-dev ? > > Chris Bryant skribis: > > Ok Stephane, that fixed it, thanks. I'm using debian unstable here, to be sure > > Stephane Fillod wrote: > > Thanks for confirming the libltdl-dev issue. On a side note, that means we have > > to fix that problem before thinking of releasing hamlib 1.2.11. > > Chris Bryant wrote: > > Btw I didn't have that probelm building 1.2.10 from source (d/l from the > > sourceforge d/l page) > > > Sorry I'm arriving late to the party here folks... > > I suspect that this libltdl problem is related to the change I made on > 2010-03-01 (r2841) which affected the way hamlib selects which libltdl > to use: > http://sourceforge.net/mailarchive/message.php?msg_name=1267470158.7216.21.camel%40fourier Kamal, I still think you did the right thing with the libltdl update. However, I don't understand why the order was changed in Makefile.am in the definition of SUBDIRS: -SUBDIRS = macros include lib libltdl src @BACKEND_LIST@ @ROT_BACKEND_LIST@ \ +SUBDIRS = macros include lib src $(subdirs) @BACKEND_LIST@ @ROT_BACKEND_LIST@ \ According to me, 'src' should be built after $(subdirs), since src depends on libltdl, when selected. Do you agree? SUBDIRS = macros include lib $(subdirs) src @BACKEND_LIST@ @ROT_BACKEND_LIST@ \ > At that time, the group discussed (but decided against) the idea of > removing the bundled libltdl from hamlib altogether, which would force > users to want to build hamlib from source to use their distro's native > libltdl (or build that too if their distro doesn't provide it). FWIW, > it sounds like Chris' problem may have been avoided if we had taken that > step. Actually, I'm not sure what was Chris' problem. Probably, it's more like me guessing it's because of libltdl-dev, while in fact the configure script was too slow finding the python dev setup. The "find /usr ..." is not that smart in macros/python.m4, and probably has to be reworked, maybe with a more up to date script. > The group may want to consider the idea of dropping the bundled libltdl > again. I'm not sure how many hamlib users even "need" it anymore (I > think libltdl is probably supplied by most distros), but we do have > evidence that its presence causes some problems: CVE-2009-3736, and now > Chris' problem. Is the benefit of bundling libltdl worth the burden > that it imposes on hamlib? Yes, it is worth it because some non-Linux distos are frugal, or carry broken library. For example, the cross-compilation for win32 under Linux does need the bundled libtldl. 73 -- Stephane - F8CFE |