|
This is the first program I've tried installing on
Unix. I seem to be having a problem similar to one submitted on June
16. I am running --disable-shared.
I was just wondering if a fix has been found, or
any suggestions at all would be helpful.
I'm trying to run htdig-3.2.0b3 on BSDI 4.2 &
GCC 295.2 with Apache 1.3.20.
Here's the message I get when I run
gmake:
/usr/lib/libc.a(ctime.o): In function
`mktime':
ctime.o(.text+0x1a90): multiple definition of `mktime' ../htlib/.libs/libht.a(mktime.o):/usr/local/htdig-3.2.0b3/htlib/mktime.c:379: first defined here ld: Warning: size of symbol `mktime' changed from 29 to 26 in ctime.o gmake[1]: *** [htdig] Error 1 gmake[1]: Leaving directory `/usr/local/htdig-3.2.0b3/htdig' gmake: *** [install-recursive] Error 1 Thank you,
Wendy
> According to Joe R. Jah: > > On Tue, 12 Jun 2001, Gilles Detillieux wrote: > > > any of the previous segfault problems you had. It seems to stem from > > > shared library constructors, so try configuring with --disable-shared > > > and see if the problem goes away. > > > > It doesn't even reach that point; I got the following compilation error:( > > _____________________________________________________________________________ > > /usr/lib/libc_p.a(ctime.po): In function `mktime': > > ctime.po(.text+0x1bdc): multiple definition of `mktime' > > ../htlib/.libs/libht.a(mktime.o):/tmp/htdig-3.2.0b4-061001/htlib/mktime.c:371: > > first defined here > > ld: Warning: size of symbol `mktime' changed from 88 to 36 in ctime.po > > gmake[1]: *** [htdig] Error 1 > > gmake[1]: Leaving directory > > `/tmp/htdig-3.2.0b4-061001/htdig' > > gmake: *** [all-recursive] Error 1 > > _____________________________________________________________________________ > Well, this is a completely new error, which you've never reported before > with 3.2 betas, and there haven't been any changes to the mktime support > in probably more than a year! Something is just not right here. Try > reinstalling the htdig-3.2.0b4-061001 snapshot completely from scratch, > to make sure your not fighting against corrupted remnants of previous > attempts. Run ./configure with the --disable-shared option, enable > profiling as you did before. Then, if you can compile, try running with > profiling again. Please use this procedure consistently with any new > snapshot you test out, so we don't end up wasting time chasing after > ghosts. > ... > > I have already tried htdig with --disable-shared on a pre-3.2.0b3 snapshot > > without any increase in indexing speed; however, I will try do profiling > > with --disable-shared, if I can get it to compile. > > Well, the fact that you did get a pre-3.2.0b3 snapshot to compile suggests > to me that you should be able to get the latest snapshot built too. > > This shared library support that Loic slapped in before running off has > been nothing but trouble, and I'd suggest that anyone experiencing the > slightest bit of oddity in the 3.2 betas avoid shared libraries like > the plague. Trying to debug this crud has been a real nightmare, and > personally, I'd like nothing better than to see it taken out entirely. > At the very least, we really ought to make --disable-shared the default, > for all systems, and give huge warnings and disclaimers to anyone who > enables it. |
- [htdig-dev] BSDi segfault revisited;( Joe R. Jah
- Re: [htdig-dev] BSDi segfault revisited;( Gilles Detillieux
- Re: [htdig-dev] BSDi segfault revisited;( Geoff Hutchison
- Re: [htdig-dev] BSDi segfault revisited;( Joe R. Jah
- Re: [htdig-dev] BSDi segfault revisited;( Gilles Detillieux
- RE: [htdig-dev] BSDi segfault revisited;( Wendy Reetz
- RE: [htdig-dev] BSDi segfault revisited;( Wendy Reetz
- Re: [htdig-dev] BSDi segfault revisited;( Gilles Detillieux
- Re: [htdig-dev] BSDi segfault revisited;( Joe R. Jah
