tech-pkg archive

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index][Old Index]

Re: The sad state of pkgsrc [was Re: pkgsrc NetBSD 7.0.1_PATCH/x86_64 2016-09-04 11:03]



On Tue, Sep 06, 2016 at 10:46:36PM +0200, Joerg Sonnenberger wrote:
 > > devel/mico                                   tonnerre%NetBSD.org@localhost
 > 
 > /data/scratch/devel/mico/work/mico/./idl/idl -B../.. --c++-skel --name CosNaming \
 >         ../../include/coss/CosNaming.idl
 > [4229412761 - 0x7f7ff6010700] address.cc:555: assertion failed

This has been discussed before and (iirc) is caused by your build
machine's hostname not resolving.

 > > devel/tcl-tclOO                              rodent%NetBSD.org@localhost
 > 
 > ./generic/tclOOMethod.c:511:14: error: incompatible types when assigning to type 'union <anonymous>' from type 'void *'
 >    cfPtr->cmd = NULL;

http://mail-index.netbsd.org/tech-pkg/2014/06/28/msg013321.html

 > > lang/cint                                    drochner%NetBSD.org@localhost
 > 
 > Invalid use of std::fpos. IMHO the package should be removed, if someone
 > wants a working C/C++ interpreter, cling is likely the better choice.

I agree on the package; it is terrible. However, I don't see any
package called "cling" :-)

 > > lang/intercal                                pkgsrc-users%NetBSD.org@localhost
 > 
 > PLIST doesn't match update (mef).

Fixed.

 > > lang/nhc98                                   kristerw%NetBSD.org@localhost
 > 
 > Various issues starting from correctly invoking the compiler.

I'm told that merging this will probably help, but many of the changes
in the patch are clearly gratuitous and I haven't had time to look
further: http://www.pfitzenmaier.de/posts/build-nhc98.html

 > > lang/smlnj                                   richards+netbsd%CS.Princeton.EDU@localhost
 > 
 > [1]   Segmentation fault (core dumped) "${BINDIR}/sml" ...
 > 
 > > lang/smlnj11072                              richards+netbsd%CS.Princeton.EDU@localhost
 > 
 > [1]   Segmentation fault (core dumped) "${BINDIR}/sml" ...

I'm aware of this (and have had to work around it, because I have
acquired a pile of legacy code that won't run in anything else) but I
have no clue how to fix it. It seems to be progressive memory
corruption, but I haven't found a cause, and as best I can tell
upstream is pretty well dead.

(smlnj11072 at least builds and runs on i386, which is why it exists.
I think the bootstrap kit in the 110.73 distfile comes already
corrupted.)

 > > math/vowpal_wabbit                           cheusov%NetBSD.org@localhost
 > 
 > parser.cc:246:7: error: 'sockaddr_in' was not declared in this scope
 >        sockaddr_in client_address;

Seems readily fixable...

 > > misc/hanzim                                  rxg%NetBSD.org@localhost
 > 
 > Needs Xserver. Does anyone actually care about this package?

I'm guessing not. It's been that way for years, and it would be pretty
straightforward to fix it up to not need the X server during build,
but nobody's bothered.

 > > security/sfs                                 pkgsrc-users%NetBSD.org@localhost
 > 
 > refcnt.h:421:18: error: declaration of 'typedef class ptr<T> ref<T>::ptr' [-fpermissive]
 >    typedef ptr<T> ptr;

That's an error?

 > > shells/standalone-tcsh                       agc%NetBSD.org@localhost
 > 
 > Doesn't honour PREFIX.

not a bug

 > > sysutils/strace                              christos%NetBSD.org@localhost
 > 
 > Needs sources.

...perhaps if there were a way to handle packages like this?

 > > textproc/hs-texmath                          szptvlfn%NetBSD.org@localhost
 > 
 > cc1: out of memory allocating 4072 bytes after a total of 0 bytes

that seems bogus...

-- 
David A. Holland
dholland%netbsd.org@localhost


Home | Main Index | Thread Index | Old Index