Re: Creating Solaris Packages

2004-11-01 Thread Niels Baggesen
On Mon, Nov 01, 2004 at 12:41:25PM -0600, Jason Frank wrote: > Tested that out too. Same problem exists, it's not passing the prefix > or exec_prefix down either: > > bash-2.05$ make install prefix=`pwd`/pkgroot/ exec_prefix=`pwd`/pkgroot/ For the internal packages that I build I use prefix

Re: MinGW Builds Failing

2004-11-01 Thread Andy Smith
Robert Story (Coders) wrote: On Mon, 01 Nov 2004 06:44:30 -0500 Andy wrote: AS> At any rate I would like for someone else to confirm or denounce the patch AS> so that we don't end up with a broken MinGW build in a release version. It turns out that the gcc cross-compiler for linux does indeed def

Re: Creating Solaris Packages

2004-11-01 Thread Jason Frank
You're absolutely right. I've used /usr/ccs/bin/make. and /usr/xpg4/bin/make, but haven't built GNU make yet. I'll get that done overnight and see if I don't get better results. Just for the record, this is not documented in README.solaris, and without it, it won't work correctly ;-) Jason

Re: Creating Solaris Packages

2004-11-01 Thread Coders
On Mon, 1 Nov 2004 11:16:00 -0600 Jason wrote: JF> I'm currently working on a "real" Solaris package for net-snmp (i,e, JF> using Sun's pkgadd and friends.) The process basically involves JF> installing to a dummy directory, building the package, then JF> substituting in the real directory for

Re: valid community strings

2004-11-01 Thread David T. Perkins
HI, Dave't answers below are right on target. One other consideration is the interoperability with managers. Some poorly designed managers make assumptions about community string values, such as ONLY ASCII letters and digits (no spaces or punctuation characters). Also, some agents also have the

Re: Creating Solaris Packages

2004-11-01 Thread Jason Frank
Tested that out too. Same problem exists, it's not passing the prefix or exec_prefix down either: bash-2.05$ make install prefix=`pwd`/pkgroot/ exec_prefix=`pwd`/pkgroot/ . . . making installlibs in /home/jfrank/net-snmp-5.1.2/snmplib mkdir /opt/net-snmp mkdir: Failed to make directory "/opt/net

RE: Creating Solaris Packages

2004-11-01 Thread Bruce Shaw
This is documented in README.solaris make install prefix=/tmp/net-snmp exec_prefix=/tmp/net-snmp >I'm currently working on a "real" Solaris package for net-snmp (i,e, >using Sun's pkgadd and friends.) The process basically involves >installing to a dummy directory, building the package, then

Creating Solaris Packages

2004-11-01 Thread Jason Frank
I'm currently working on a "real" Solaris package for net-snmp (i,e, using Sun's pkgadd and friends.) The process basically involves installing to a dummy directory, building the package, then substituting in the real directory for the dummy one inside of the newly created package. So, I'm do

Re: MinGW Builds Failing

2004-11-01 Thread Coders
On Mon, 01 Nov 2004 06:44:30 -0500 Andy wrote: AS> At any rate I would like for someone else to confirm or denounce the patch AS> so that we don't end up with a broken MinGW build in a release version. It turns out that the gcc cross-compiler for linux does indeed define WIN32. I can only assume

Re: MinGW Builds Failing

2004-11-01 Thread Andy Smith
Robert Story (Coders) wrote: On Fri, 29 Oct 2004 22:51:53 -0500 Andy wrote: AS> Robert Story (Coders) wrote: AS> but in mingw32msvc.h I define mingw32 for consistency [...]. But as for AS> the patch #ifndef WIN32 works. Are you sure? Because the module processing code doesn't include the net-snmp

Re: MinGW Builds Failing

2004-11-01 Thread Andy Smith
Interesting. I was under the assumption that the build environment added it at compile time. I know that you have to add WIN32 explicitly with Cygwin, or at least that has been my experience. I have never delved too deeply into the internals of MinGW, but I do know that using WIN32 in #ifdef bl

Re: valid community strings

2004-11-01 Thread Dave Shield
> I would also like to know whether community string can be unicode > or is it only ASCII? The same answer applies: > Nothing inherent in SNMP, no. > Assorted MIB tables make various assumptions DisplayString is defined as being ASCII only. SnmpAdminString can take UTF values. So if you used

Re: sub-agent SIGPIPE?

2004-11-01 Thread Dave Shield
RS> I was thinking of moving the signal registration, for the signals which RS> are ignored, into one of the init routines. RS> Does that sound reasonable? It is a change in behavior for the agent RS> library. Though there is a good argument for it, as it does RS> cause a crash for a not-unr

Re: MinGW Builds Failing

2004-11-01 Thread Dave Shield
> The attached patch fixes the MinGW build. Thanks Robert! I realize that > MinGW may not be a show stopper to most, so we can hold off or commit. > Has the rc been cut yet? I don't know whether it had been cut at the time you asked the question. but it's certainly out now. Given the discussio