This is a log of the Netatalk Release Hacking Party at Saturday, 2001-11-24 on #netatalk-devel@irc.openprojects.net. I have removed all personal chatter and generally everything, not directly related to Netatalk. I've also anonymized personal data (like IP addresses). Participants: The nicks have partly changed during the course of the chat, but they should still be recognizable. * Dan: Daniel E. Lautenschleger * JamieFlournoy: Jamie Flournoy * PaulKrohn: Paul Krohn * Steve: Steve Freitas * arichard: Alain Richard * bs_Tru64: Burkhard Schmidt * craigz: Craig Zimmerman * dnanetatalk: Dallas Antley * jroger: Sebastian Rittau * lancel: Lance Levsen * marcus: Joe Clarke * schluppy: * sibaz: Simon Bazley * stewb: Stew Benedict * swanberg: Karen A Swanberg --> jroger (~srittau@jroger.in-berlin.de) has joined #netatalk-devel --- Topic for #netatalk-devel is Netatalk Release Hacking Meeting today 17:00 UTC -- logs are kept -- test tar balls available at http://me.in-berlin.de/~jroger/netatalk/ --- Topic for #netatalk-devel set by jroger at Sat Nov 24 16:31:58 --- ChanServ gives channel operator status to jroger --> Dan_SuSE (~usrdel@144.92.19.115) has joined #netatalk-devel --> swanberg (swanberg@Jolinar.geo.umn.edu) has joined #netatalk-devel --> HeliumGas (~nobody@cl3029622-b.rchstr1.mn.home.com) has joined #netatalk-devel --> stuart (~stuart@dsl254-099-164.nyc1.dsl.speakeasy.net) has joined #netatalk-devel --- jroger has changed the topic to: Netatalk Release Hacking Meeting -- logs are kept -- test tar balls available at http://me.in-berlin.de/~jroger/netatalk/ Karen, what type of computer lab to you maintain? --- Lance_Away is now known as lancel Anyone here got NT to compile on an OS X? * swanberg grins at Lance. "I presume NT here means Netatalk, and not WinNT? =)" :) AFAI heard, there are still issues with OSX. I tried to resolve them, but I didn't manage, since I don't have access to an OSX box. Mine fails on the ./autogen.sh, it can't find a couple of macros. I've only used OSX 10.1.1 as a client for Ntatalk. --> marcus (~marcus@rdu57-28-046.nc.rr.com) has joined #netatalk-devel I'm afraid I'm only going to be able to test OBSD 3.0 today. My 2.9 box is busy ghosting my WinNT server... (I love dual boots...) No sweat, Karen. So, where's the rc source? I notice a warning during compilation that in afpd, it says dangerous to use mktemp, better to use mkstemp (or something like that). This a problem? lancel: I'm just uploading rc0.3. --> arichard (~Snak@ns1.equation.fr) has joined #netatalk-devel me.in-berlin.de/~jroger/netatalk/ Done. Is rc.3 the one we're all gonna start with? I would suggest so. rc0.1 was release yesterday or so and had a few problem reports already. In rc0.2 I (hope I) fixed the problem reported by Steve: libdb3 got linked to every binary. In rc0.3 I fixed the Tru64 problem reported by Burkhard Okay. Thanks. Sebastian, are you logging this? Dan_Suse, no, the mkstemp is not required here. We don't actually open the temp file. Yep. Thanks, Joe. <-- jroger has kicked [jroger] from #netatalk-devel (Daemon be gone.) --> JamieFlournoy (~chatzilla@adsl-63-204-195-98.dsl.snfc21.pacbell.net) has joined #netatalk-devel --> steve (~steve@cx428818-b.irvn1.occa.home.com) has joined #netatalk-devel <-- JamieFlournoy has quit (Remote closed the connection) --- steve is now known as SteveF --> JamieFlournoy_ (~chatzilla@adsl-63-204-195-98.dsl.snfc21.pacbell.net) has joined #netatalk-devel lancel: I will strip out all stuff not directly related to netatalk and post the log on netatalk-devel if that's fine with everyone. --> sibaz (~root@sibaz.ncuk.net) has joined #netatalk-devel Sounds good! hi guys Simon? fine indeed --> PaulKrohn (~pkrohn@agency.sfna.com) has joined #netatalk-devel Simon Bazley, aka sibaz * JamieFlournoy_ getting-tarball openbsd: I ran the 0.1 a few minutes ago, ./configured, and then deleted that dir to run 0.3. On 0.3, now, I get this error: "checking whether build environment is sane...configure: error: newly created file is older than distributed file!" Karen, sounds like either you're building on a NFS volume, or your time is really off sync. yup, my time was off. now it's working... Any word on whether jeff was planning to be here? SteveF: Hadn't heard him say anything about this meeting. not sure? is Joe here? sibaz: I'm here. ./configure --prefix=/usr/local/netatalk --sysconfdir=/etc --enable-suse --enable-timelord --with-pam --with-shadow --with-flock-locks --with-tcp-wrappers --with-did=last --with-pkgconfdir=/etc/atalk works just fine! also my first time using cvs ... what a bear! Cause I wanted to respond to Joes comments about the logger code going into the main trunk of cvs are we checking just for compile success, or running success? Please, respond. cvs is lovely *beams* PaulKrohn: You don't need to use CVS. Just use the supplied tar-balls. Karen: Both! :-) In fact Paul, please stick to the tar-balls, otherwise it's more of a moving target. Good build on Debian w/ ./configure --prefix=/home/lance/netatalk_build --with-tcp-wrappers --with-shadow so whats the plan and who's chairing this meeting? sibaz is Sebastian, correct? * jroger shakes his head. I'm Sebastian. ok, tarballs are a-fetching ... nope, sibaz is simon bazley, aka the one thats written the logger and is using a linux 7.1 box in london (sibaz si mon baz ley) --- SteveF is now known as Steve_Mandrake sibaz: I will post the plan I had posted on the mailing list at the URL in the topic. Do we need a chair? --> stewb (~stewb@209.81.166.20) has joined #netatalk-devel I think "~/ "Home Directory" is AppleVolumes.default should be left as #~ with a note mentioning that home directories can be enabled by removing the pound sign. it would probably make things easier, but I not sure its really a necesity oh, okay, As long as you're here. =) How does the rc.atalk file work now? It gets generated, and we just have to refer to it in rc or rc.local somewhere? on RH 7.2 linux, a simple ./configure seams to compile ok, but there are some errors on what is going installed --> craigz (~craigz@66-108-57-38.nyc.rr.com) has joined #netatalk-devel Dan, I disagree. I think when an admin installs it, it ought to work to some degree out of the box. That way they don't freak when they run it. Yes, it could be a matter of RTFM, but not everyone does. Hello. I'm Craig Zimmerman, looking at building for OSX. Hi Craig! Thanks for coming! * swanberg grins at the brave OSX man... Any other opinions? env CFLAGS="-O -pipe" ./configure --without-pam --with-pkgconfdir=/usr/local/etc --with-did=last --with-tcp-wrappers works on FreeBSD. Having trouble with afp_options.c craigz: what's the error? You are assuming that an admin wishes to share Home directories, though... Dan, that's true, I just think something ought to work by default. Are there other options? (...of what could be shared?) craigz: I have heard there were problems on OSX. Related to macro concatenation. I had traced down this problem to include/atalk/paths.h Dan_SuSE: Can't the user just create their own .applevolumes file, and share out their home dirs? I don't know, Joe. marcus-> here goes: afp_options.c: In function `afp_options_init': afp_options.c:132: `_PATH_LOCKDIRafpd' undeclared (first use in this function) afp_options.c:132: (Each undeclared identifier is reported only once afp_options.c:132: for each function it appears in.) afp_options.c: In function `afp_options_parse': afp_options.c:366: warning: comparison is always false due to limited range of data type how about a default share of a directory within PREFIX? would at least serve as an example. you could ship it with a dummy volume shared, with a text file in it saying "OK now customize your shared volumes by editing..." etc. I've just edited paths.h to have the line: define _PATH_LOCKDIR "/var/run/" Rather than looking for the #IFDEF MACOSX_SERVER I like anything other than having something shared by default. Sebastian: Compiling RC0.3 on Mandrake 8.1 with CNID. I took DB3 out of ldconfig's cache. Compile, make, make install worked file. When I attempt to start the daemon, I get this: Starting AppleTalk services: *I* want to chosse what is to be shared. Starting atalkd:/usr/local/netatalk/sbin/atalkd: error while loading shared libraries: libdb-3.3.so: cannot open shared object file: No such file or directory [FAILED] Registering schizo:Workstation: /usr/local/netatalk/bin/nbprgstr: error while loading shared libraries: libdb-3.3.so: cannot open shared object file: No such file or directory [FAILED] Registering schizo:netatalk: /usr/local/netatalk/bin/nbprgstr: error while loading shared libraries: libdb-3.3.so: cannot open shared object file: No such file or directory [FAILED] Starting papd:/usr/local/netatalk/sbin/papd: error while loading shared libraries: libdb-3.3.so: cannot open shared object file: No such file or directory [FAILED] Starting afpd:/usr/local/netatalk/sbin/afpd: error while loading shared libraries: libdb-3.3.so: cannot open shared object file: No such file or directory [FAILED] JamieFlournoy_: I like your idea of a non-writable directory with a SimpleText file saying, "Congratulations! You rawk! Now customize $CONFIG_PATH/AppleVolumes.system! craigz: The problem is that obviously the wrong concatenation rule ATALKPATCAT in paths.h is used. --- JamieFlournoy_ is now known as JamieFlournoy_Solaris Steve_Mandrake: I will look into this. Could you send me the relevant log part via /dcc send? jroger any ideas what ATALKPATCAT concatenation rule to use? craigz: If you look at paths.h you will see that there is a conditional #define of ATALKPATHCAT. Could you find out which one is used? Possibly using #error. --- swanberg is now known as swanberg_obsd For some reason the startup script for SuSE is ending up back in /sbin/init.d/ It should be going to --sysconfdir There seems to be a hardcoded path for atalkd.tmp somewhere. Using --prefix=/home/lance/netatalk_build, syslogs outputs a "/usr/local/netatalk/etc/netatalk/atalkd.tmp: No such file or directory" jroger -> here's the part from paths.h: #ifdef HAVE_BROKEN_CPP #define BROKEN_ECHO(a) a #define ATALKPATHCAT(a,b) BROKEN_ECHO(a)##BROKEN_ECHO(b) #else #define ATALKPATHCAT(a,b) a##b Dan_SuSE: That location is hard-coded in the Makefile. #endif #else #define ATALKPATHCAT(a,b) a/**/b #endif Where can I define HAVE_BROKEN_CPP, that is OSX Like. i have the console windo open, but it just shows the welcome info, never what i type. hi everybody, it seems I have missed the start of meeting? I have just discovered there is a rc0.3 version I have just downloaded as an asside, are we going to do anythign about atalkd acting as a router? At the very least it shouldn't try to be a router unless there are 2 network cards, alternatively it shouldn't try atall unless its told to. it there any recommended ./configure options ? sibaz: I would suggest delaying that to the next release. sysvdir should be /etc/rc.d/ for SuSE Dan_SuSE: is it okay to hardcode that path? Dan_SuSE: RedHat seems to have /etc/rc.d/init.d I don't understand as I changed that in the CVS version on Monday. can someone give me an ftp address where I can download the tarball (please) jroger I put a hard coded DEFINE BROKEN_CPP in paths.h and it built a bit further, giving only a warning: afp_options.c:132: warning: cannot merge ) with BROKEN_ECHO, result is not a valid token afp_options.c: In function `afp_options_parse': afp_options.c:366: warning: comparison is always false due to limited range of data type but then croaks on quota.c as such: quota.c:408: only 1 arg to macro 'dbtob' (2 expected) quota.c:409: only 1 arg to macro 'dbtob' (2 expected) quota.c:410: only 1 arg to macro 'dbtob' (2 expected) quota.c:412: only 1 arg to macro 'dbtob' (2 expected) quota.c:413: only 1 arg to macro 'dbtob' (2 expected) quota.c:414: only 1 arg to macro 'dbtob' (2 expected) redhat use /etc/rc.d/init.d up to versions 6.2 and now recommend /etc/init.d sibaz: Doesn't http work? SuSE 7.1 -7.2 now use /etc/rc.d I could use http, but I'm inside a proxy server, ftp is easier ...also 7.3 Seb, perhaps your tarball does not have my Monday change to distrib/initscripts/Makefile.am Dan_SuSE: I'm taking the rc0 tar-balls out of the CVS release-branch so not all changes may have propagated. Dan_SuSE: Yes... Dan_SuSE: Let's see... Nope, it does not...line 48 craigz: try defining HAVE_BROKEN_DBTOB Dan_SuSE: You're using rc0.3? I had changed it since rc0.1. I just ripped 2GB of data to a shared Netatalk volume and an NT volume. Netatalk was almost twice as fast!! Yes, rc.3 Dan_SuSE: Netatalk's breath stinks with the blood of man. *giggles* Uh, yeah, Steve... You're scaring me. Dan_SuSE: Hahahaha.... Was NT serving DDP or TCP? NT only supports AppleTalk, not AppleTalk over IP. Dan_SuSE: In my version it says in line 48: :SBINDIR:/timelord Oh. Well. That would explain it, then. :-) Nah, I was just curious whether you were comparing to MacServerIP on NT or the like... * swanberg_obsd growls that some people abbreviate NetaTalk to NT, and now she's all confused... * Steve_Mandrake growls that people USE NT... * lancel Agrees w/ steve No, look at Makefile.am. It is missing my Monday change on line 48. marcus, I defined HAVE_BROKEN_DBTOB in quota.c, and now get a different error: quota.c:376: undefined or invalid # directive quota.c:419: #endif outside an #if There is a problem with afpd finding the uams on a --prefix=/not/usr/local build. craigz: You you send me your config.h file to srittau@debian.org? jroger it's on the way.... Could everyone who has the compile, install, and Netatalk function working on their platform please yell? And say what platform, please. (Without issues, that is.) --> schluppy (~sschluppe@dialin-145-254-032-009.arcor-ip.net) has joined #netatalk-devel can someone tell me where I can get a copy of the tarball from to test please sibaz: http://me.in-berlin.de/~jroger/netatalk/ * Steve_Mandrake yells, obeying his own voice, saying Mandrake works great without db3 in the path. ok, i compile with very simple options to start, just defining prexif and sysconfidr. on solaris 2.6, i hadn't installed the openssl stuff, and that worked, at least to the extent of seeing my home directory,. rats. remember the problem OpenBSD was having with not creating the .so files? Still happening! craigz: You might try disabling quota support by setting NO_QUOTA_SUUPORT. Thanks Paul whats the full url to the file (I'm using wget to download it) SuSE 7.3 is working fine except rc.3 needs to have distrib/initscripts/Makefile.am line 50 modified to read: sysvdir = /etc/rc.d Not sure why that #endif is out side of range. Of course, the final few #if..#endif's were not commented. on solaris 2.8, i have openssl installed, and noow i think i'm in PAM hell ... i mean heck! http://me.in-berlin.de/~jroger/netatalk/netatalk-1.5rc0.3.tar.gz Dan_SuSE: I'll take your word for it that'll be fixed. :-) cheers thats only 754k is that right? yeah I mean 764k err, yeah, that too Dan_SuSE: done so for rc0.4 The Obsd problems with .so files were once upon a time due to libtool issues, right? I don't have libtool on this machine yet, didn't think I'd need it. I compiled with --with-tcp-wrappers --with-ssl-dir=/usr Thanks, Seb. 764590 bytes. --- swanberg_obsd is now known as swan_obsd Stew, how goes Solaris 7? Alain, had a chance to dig into YellowDog yet? One question, is there a 'make uninstall' feature that will remnove an installed Netatalk app? I'm tired of manually uninstalling Netatalk. Steve - just running a straight .configure right now with NO_QUOTA_SUPPORT I get: quota.c:47: undefined type, found `DEFINE' quota.c:47: illegal external declaration, missing `;' after `NO_QUOTA_SUPPORT' quota.c:376: undefined or invalid # directive quota.c:380: #endif outside an #if quota.c:419: #endif outside an #if How are you defining it, craigz? Dan_SuSE: There is, but I don't know if it works. It would be a bug if it doesn't, though. I mean, this kind of thing should be in config.h think that was a typo...just made it: #define NO_QUOTA_SUPPORT #define NO_QUOTA_SUPPORT 1 in config.h should do it for testing. Cool Schluppy, what platform are you compiling on? I've done a successful configure and make on the targball (running on redhat linux 7.1) Steve_Mandrake We use SuSE 7.X and SuSE Enterprice Server 7 Cool. * JamieFlournoy_Solaris compiling on Solaris 8 can anyone access a macserver on the internet via IP if I turn it on? Steve_Mandrake Sumit Bose from SuSE is the person who help us to get a version to work, he has posted some mails in the mailling list I can, Simon. under RH 7.2, the rc0.3 version build with ./configure --prefix=/opt/netatalk --enable-redhat compiles and install without problems. I am now trying it with a MacOS 9 client and I have a problem : during the copy of a full system folder (about 260 Mo and 2000 files), the copy process stop at half with an error "unable to copy QuickHelp� because an item with the same name already exits". Alain, you should also be using the --with-did=last option to configure. Does anyone know what controls the libdir in ./etc/uams/uams_*. No thoughts on the lack of links to .so files? Marcus, I've edited config.h and it's building further now.... Ok, I'll make install and put a quick AppleDefaults file together. I'm doing a strictly basic install wiht no configure options as I reccon thats what most basic users will do marcus: That's the default for rc's. marcus: ok, I recompile (I was assuming this was the default option). Sebastian, is --with-did=last the default now? Alain should have gotten that without having to specify anything, I think. oh yeah. A recurring Solaris "make" issue that could be fixed with docs: "ar" isn't in a user's PATH by default; it's in /usr/xpg4/bin and so it needs to be added or else "make" will fail. I'll set up an account for you Steve, password Mandrake (with capitals) Karen: Do you know what causes this? swan_obsd: What does 'find . -name "*.so"' say? ok, having trouble with pam.conf on solaris 8 ... i added the lines from SYSCONFDIR/pam.d/netatalk to /etc/pam.conf, having added 'netatalk' to the start of each line if netatalk is the user's first compile from source on Solaris, then the user probably won't know this off the top of their head. steve, my box is dns sibaz.ncuk.net ip 213.166.20.162 Steve_Mandrake: It is. JamieFlournoy_Solaris: Looking into that... Simon: Got it, just tell me when you want me to connect. Alain, do you have USE_LASTDID defined to 1 in config.h? * JamieFlournoy_Solaris make-installing ok, I've done a make install, I'll just setup the appledouble and run it and tell you default install hasn't generated the atalkd scripts in init.d marcus: I have just checked and I have effectively USE_LASTDID=1 in config.h, so this is not a config problem but a run problem. Also retrying the same copy result in the same problem. The only special thing I have is "codepage:maccode.iso8859-1" in the AppleVolumes.default file, so I will make a try without it steve: If I knew what caused it, I'd fix it. =) This has been a persistant problem, check out the netatalk-devel archives, thread: uams still broken BSD from July. sebastian: where should I do that find for the .so files? In usr or in my compile directory? swan_obsd: Sorry, I was just trying to kickstart the exchange for you, but Seb jumped in. :-) I'll try compiling without the --with-ssl-dir/usr. steve, I'm just trying to find the binarys now *grins* swan_obsd: I think we have the same problem. Does your syslog say it can't find the uams? swan_obsd: The compile directory (netatalk-1.5rc0.3) i get "User unknown or incorrect password ..." when trying to connect to netatalk on Solaris 8 w/ssl. Also, are we expecting ddp to work on Solaris 8? perhaps only in 32-bit mode? marcus: no, i just use an empty folder in which I copy my system folder. I manually uninstalled Netatalk rc.3, removed did and used cnid. All compiles properly on SuSE and appears to function properly as well. swan_obsd: I think the problem is in ./etc/uams/uams*.la where the libdir is hardcoded to libdir='/usr/local/netatalk/etc/netatalk/uams' this isn't where the uams are if you used --prefix jroger: The find command doens't find anything in the compile directory Ok, I will make up a list of all problems that people have with rc0.3 Please do "/msg jroger [short problem desc]" with your problems. Nother error: 'There is not enough room on the disk "Home Directory" to copy (an additional 278,478,848 GB is needed).' This on the Solaris 2.6 i thought was working ... marcus: there is only one quickhelp� file in that folder ! swan_obsd: Can you post the output of "make" while it's going through the etc/uams directory? OK, I've found a problem, the rc.atalk.redhat file in distrib/initscripts isn't executable any ideas which ports I need to openup on IP chains for this to work? sibaz: 548 tcp is afpovertcp. infact, steve whats your IP addres I'll just give you complete access XX.XXX.XX.XXX Okay, the list is now up at http://me.in-berlin.de/~jroger/netatalk/compile-problems.lst . I will add/remove from this list as new problems are discovered/problems are resolved. Please /msg me with additions. --> opti (~bs@cube.cpfs.mpg.de) has joined #netatalk-devel marcus: i just restart the copy without the codepage:maccode.iso8859-1 option and now the copy go up to 2/3 but now hangs with an other problem : "error -37" during the copy of "Ressources syst�me". In the MacErrors.h file, the error is defined as "bdNamErr = -37 /*there may be no bad names in the final system!*/" so this is not supposed to occure ! marcus: I will make more investigation with --enable-debug --> sibaz2 (~root@sibaz.ncuk.net) has joined #netatalk-devel arse, my connection to my server just dropped I'm ready for you now steve if you want to try it error -37 is bdNamErr Bad file name; there may be no bad names in the final system! <-- sibaz has quit (Remote closed the connection) This -37 error is the same error that occurs when copying files with OS X clients. steve, any joy? sibaz2: Getting incorrect password error. I'm using "Steve" and "Mandrake". MANDRAKE all caps thats right <-- opti (~bs@cube.cpfs.mpg.de) has left #netatalk-devel OK, I've had this problem before but I thought it was fixed I'm using shadow passwords, and it used to not detect it properly --> opti (~bs@cube.cpfs.mpg.de) has joined #netatalk-devel my syslog message is this: Hmm, I tried "steve" and "Mandrake" and now you've hung my Mac! I'm thinking civil suit. LOL jroger: How do you want us do corrections? CVS to sourceforge? Nov 24 17:59:51 sibaz afpd[2110]: ASIP session:548(1) from 24.177.19.184:62098(2) Nov 24 17:59:51 sibaz afpd[2110]: dhx login: Steve Nov 24 17:59:52 sibaz afpd[2110]: uams_dhx_pam.c :PAM: PAM_Error: Authentication failure -- No such file or directory Nov 24 17:59:52 sibaz afpd[2110]: 0.17KB read, 0.13KB written Nov 24 17:59:52 sibaz afpd[2014]: server_child[1] 2110 done Nov 24 18:00:17 sibaz afpd[2115]: ASIP session:548(1) from 24.177.19.184:62098(2) Nov 24 18:00:18 sibaz afpd[2115]: dhx login: Steve marcus: so I just retry with iso conversions and debug. I was thinking that most people where not using codepage:maccode.iso8859-1 option although I think this should be a default option. Nov 24 18:00:18 sibaz afpd[2115]: uams_dhx_pam.c :PAM: PAM_Error: Authentication failure -- No such file or directory Nov 24 18:00:18 sibaz afpd[2115]: 0.17KB read, 0.13KB written Nov 24 18:00:18 sibaz afpd[2014]: server_child[1] 2115 done sorry steve *hugs* Do wipe my tears. *wipes Steves tears, but mutters somehing about the wonder reliability of macs* sibaz-> can you tell us what ls -l is of your /etc/uams directory? lancel: Patches that I can apply to my local copy of the stablizing branch and roll into the RC's. When we're finished today, I will commit all these patches. no such file or directory Sibaz, where are the UAMs? Do an ls -l of that directory. ...yes, that's what I meant. I've not got the foggiest, whats the default cause thats where they are (I'm doing a find now) but somethings wrong, like I say I'm being a complete simpleton with my setup, whatever its doing, its wrong if it hasn't worked its own thing out just seen that there is a maccode.iso8859-1.adapted file. Who knows what it is ? ok, /usr/local/etc/netatalk/uams Sibaz, what distro are you using and what is your ./configure line? I just typed ./configure no options ls -l of that uam directory and post it. --> satan (ali@frank.gwc.org.uk) has joined #netatalk-devel Well, that may explain the problem. here you go: lrwxrwxrwx 1 root root 11 Nov 24 17:45 uams_clrtxt.so -> uams_pam.so -rw-r--r-- 1 root root 48778 Nov 24 17:45 uams_dhx_pam.a -rwxr-xr-x 1 root root 869 Nov 24 17:45 uams_dhx_pam.la -rwxr-xr-x 1 root root 50264 Nov 24 17:45 uams_dhx_pam.so -rw-r--r-- 1 root root 36884 Nov 24 17:45 uams_dhx_passwd.a -rwxr-xr-x 1 root root 884 Nov 24 17:45 uams_dhx_passwd.la -rwxr-xr-x 1 root root 39325 Nov 24 17:45 uams_dhx_passwd.so * swan_obsd is back temporarily. She hate WinNT so much it makes her see red. "Sebasitan, I'm re-making so I'll post the output for the uams section in a second... lrwxrwxrwx 1 root root 15 Nov 24 17:45 uams_dhx.so -> uams_dhx_pam.so -rw-r--r-- 1 root root 25778 Nov 24 17:45 uams_guest.a -rwxr-xr-x 1 root root 822 Nov 24 17:45 uams_guest.la -rwxr-xr-x 1 root root 28660 Nov 24 17:45 uams_guest.so -rw-r--r-- 1 root root 38202 Nov 24 17:45 uams_pam.a -rwxr-xr-x 1 root root 814 Nov 24 17:45 uams_pam.la -rwxr-xr-x 1 root root 40281 Nov 24 17:45 uams_pam.so -rw-r--r-- 1 root root 31198 Nov 24 17:45 uams_passwd.a -rwxr-xr-x 1 root root 829 Nov 24 17:45 uams_passwd.la -rwxr-xr-x 1 root root 33892 Nov 24 17:45 uams_passwd.so -rw-r--r-- 1 root root 42354 Nov 24 17:45 uams_randnum.a -rwxr-xr-x 1 root root 863 Nov 24 17:45 uams_randnum.la -rwxr-xr-x 1 root root 43992 Nov 24 17:45 uams_randnum.so so uams)dhx_pam is there, I wonder why it didn't find it? swan_obsd: thanks I'm using redhat 7.1 --- sibaz2 is now known as sibaz2-redhat-7 --- sibaz2-redhat-7 is now known as sibaz-redhat-7-1 any ideas? Geez, I remember this was happening last week to Jon Newman. I could swear there was a patch applied to CVS and he was able to fix it. ...let me scout the devel list. BRB lancel: Regarding your hardcoding problem: I can't reproduce that here. My *.la file has libdir='/home/srittau/Test/etc/netatalk/uams' if I use prefix=/home/srittau/Test. Which other flags did you use? <-- opti (~bs@cube.cpfs.mpg.de) has left #netatalk-devel jroger: ./configure --prefix=/home/lance/netatalk_build --with-tcp-wrappers --with-shadow lancel: What platform? jroger: Debian testing configure didn't enable shadow support, I'm assuming pam should deal with that? Sibaz. I can'r find the e-mail about this issue, but I'd swear this was fixed in CVS. <-- schluppy has quit ("i will be back later and listen to you") ok dan, any idea when and I'll chec the cvs logs output of some .configure --with-tcp-wrappers --with-ssl-dir=usr on obsd: checking for SSL... /usr (enabling RANDNUM and DHX support) checking for main in -lcrypto... yes checking for main in -ldes... yes "make"ing now... lancel: Uhm... I'm using Debian unstable, so the difference shouldn't be that big... Do you have libtool installed? What version? --> bs_Tru64 (~bs@cube.cpfs.mpg.de) has joined #netatalk-devel Around Nov 16 or 17th. jroger: ltmain.sh (GNU libtool) 1.3.5 (1.385.2.206 2000/05/27 11:12:27) Post to the list and see if Jon Newman can help. ...I know he was able to resolve is Red Hat UAM issue. marcus: I have just checked the differences between maccode.iso8859-1 and maccode.iso8859-1.adapted and it seems the later has about 13 more char equivalents between iso1 and macos charset. One of them is code 0xAA, the code for �, so probably the problem is in the maccode.iso8859-1 table. I make an other try with maccode.iso8859-1.adapted table. Sibaz and Dan: So, this isn't caused by the fact that Simon didn't do --with-shadow? Sibaz, post a message to seb: /msg jroger [problem] I don'r know, Steve? Simon, please try recompiling with --with-shadow, and let's see what it does. Good idea... Steve, I know that will work, but its not a solution its a work around marcus: I have probably to first recompile without --debug because this is too slow ! (4 minutes to copy without, 7 hours with it !) Oh. What's your goal? Steve_Mandrake: build/install OK Solaris2.7 - startup fails on can't establish STREAMS plumbing, exiting. lancel: You you re-configure, re-run make and post the make output in /etc/uam, too? I always specify --with-pam and --with-shadow, Sibaz. Stew, can you /msg that to jroger? He'll add it to the list. Thanks. jroger: I'm also going to use --with-pam Here's my full configure line:./configure --prefix=/usr/local/netatalk --sysconfdir=/etc --enable-suse --enable-timelord --with-pam --with-shadow --with-flock-locks --with-tcp-wrappers --with-db3=/usr/lib --enable-cnid-db --with-pkgconfdir=/etc/atalk output of obsd make: afp_options.c: In function `afp_options_parse': afp_options.c:366: warning: comparison is always false due to limited range of data type ... Steve thats what I used to do to get this to work back on version 1.5pre6 sebastian, can I e-mail you the make output on openbsd? I'm not sure what you want to see. swan_obsd: yes, I swan_obsd: yes, I'm not sure, either, but then... Stew, just a shot in the dark, but I heard something about Solaris & DDP kernel stuff. I wonder what would happen if you went into netatalk.conf and turned atalkd and papd off. marcus: to be more effective, I may enable debug only on some part of the process, do you have an idea ? (it should be better for us to set the debug a runtime option that may be more or less verbose like all the other daemon works) Karen, can you try the following on line 365: burkhard: did you try rc0.3? Steve - I'm seeing now in the readme - I did not do make kinstall - which fails options->umask = (mode_t)strtol(optarg, &tmp, 8); make kinstall fails? Sebastian, I did. The UAMs are named correctly bash-2.03# make kinstall make: Fatal error: Don't know how to make target `kinstall' whats the CVS version of the /netatalk/configure.in file? bs_Tru64: Great, thanks. second that, no kinstall target. :( marcus: just for you information, I have reproduced the problem by copying just a part of the system folder and it reproduced on the same file, so this is probably the problem I was expecting. The other problem is that there is nothing interesting in the debug log, only the following : marcus: yes, but in a minute. I'm multitasking today... Burkhard: Since you joined late, you should know that Sebastian's keeping a list of outstanding issues and put it up for viewing at "http://me.in-berlin.de/~jroger/netatalk/compile-problems.lst". So for every problem you have that doesn't get quickly solved, please /msg jroger [short problem description] These log messages look hopeful: Nov 24 19:19:26 julie afpd[10608]: begin afp_getfildirparams: Nov 24 19:19:26 julie afpd[10608]: end afp_getfildirparams: Nov 24 19:19:26 julie afpd[10608]: begin afp_createfile: Nov 24 19:19:26 julie afpd[10608]: begin afp_getfildirparams: Nov 24 19:19:26 julie afpd[10608]: begin getfilparams: Include $(top_srcdir)/include before everything else.