- 2008-Jun-03 03:36 anonymous
- 2008-Mar-14 03:14 anonymous
- 2007-Nov-06 12:38 ono
- 2007-Nov-06 12:18 ono
- 2007-Nov-06 11:07 ono
- 2007-Nov-05 18:25 ono
- 2007-Nov-05 18:22 ono
- 2007-Nov-05 18:03 ono
- 2006-Jun-02 17:11 anonymous
- 2006-May-26 15:03 drh
- 2006-May-26 15:03 drh
- 2006-May-26 15:02 drh
- 2006-Feb-08 04:14 anonymous
- 2006-Feb-08 03:51 anonymous
- 2004-Jun-04 06:05 anonymous
- 2004-May-29 16:16 anonymous
- 2004-May-29 16:15 anonymous
- 2004-Feb-06 16:16 anonymous
- 2003-Nov-23 12:41 anonymous
- 2003-Jul-17 18:19 anonymous
- 2003-Feb-14 20:40 anonymous
- 2002-Sep-03 11:47 anonymous
- 2002-Aug-05 23:40 anonymous
- 2002-Aug-05 18:43 anonymous
cygwin install target: d:\services\cygwin
cvs repository: d:\cvs (mounted as /home/cvs inside the cygwin environment)
I setup cygwin to use the DOS mode (which I believe mounts the drives in textmode by default)
install cygwin packages:
- devel/cvs, rcs
add d:\services\cygwin\bin to all users path
I mounted my cvs dir under the cygwin root -- I believe it required a reboot to really take hold.
mount -f -s -t "d:/cvs" "/home/cvs"
An ls of /home does not show 'cvs', however you can cd into it and run commands through it.
build sqlite and cvstrac (I actually used the cygwin gcc setup to do this. Required cygwin packages: binutils, make, gcc, etc.) I used the latest CVSTrac source (1.1.0 from CVS) and SQLite source (2.6.2 tar ball), compiled under the cygwin environment using gcc and other normal binutils.
Richard mentioned that sqlite would not compile under cygwin properly, however I gave it a shot and it seems to run fine. I've tested nearly every function I can think of (that read/write from the .db files) and all appears fine. There may be considerations of size limitations that I do not know about -- Richard could comment here if possible?
NOTE: turned off I8N (Changed "define CVSTRAC_I18N" to "0" in config.h) defines because of this error: ../cvstrac-1.1.0/config.h:40: langinfo.h: No such file or directory
Move the CVSTrac binary into directory from which it will run:
cp cvstrac.exe from build dir to /usr/local/bin/cvstrac-1.1.0.exe ln -s /usr/local/bin/cvstrac-1.1.0.exe /usr/local/bin/cvstrac.exe
run the CVSTrac initialization command:
/usr/local/bin/cvstrac init /home/cvs Project-name
NOTE: most of the problems I ran into getting it all working, were path issues. The server I'm using for this also runs ViewCVS and cvsntserver. I spent a good deal of time isolating the versions of the various cvs/rcs utils (rcsdiff, co, etc.). CVSTrac now uses the versions that come w/ the cygwin packages. ViewCVS is configured in it's conf to use the same path. CVSntserver uses its own root.
In the end, I wrote an sh script that sets the path then runs cvstrac:
PATH=/bin:/usr/bin:/usr/local/bin export PATH /usr/local/bin/cvstrac cgi /home/cvs Project-Name
I ended up installing the latest apache package from within cygwin's setup util. This installs into /var/www, /etc/apache and uses /usr as its prefix dir. (Module installed: web/apache)
- Modified /etc/apache/httpd to:
- use 'index.cgi' as an index document
- set ExecCGI in DocumentRoot Directory directive
- use a new port number (other httpd's running on this host)
- proper fqdn for server name
- proper 'webmaster' email address
- Started apache
- Linked exec_cvstrac.sh to /usr/www/htdocs/index.cgi
ln -s /usr/local/bin/exec_cvstrac.sh /usr/www/htdocs/index.cgi
Visited http://MyHostName:PortNum to login to my new CVSTrac server
Initially I attemted to get this running using the cygwin Inetd port and cvstrac in 'http' mode. I was able to get the server to limp along, however it behaved badly in the following ways:
- In MS IE 5.5sp2 & 6.0 the pages would continually reload themselves over and over again. You could click the stop button, then do a CVSTrac operation. I looked through the source and played w/ changing the linefeed/carriagereturn setup for the HTTP responses, however was unable to affect a real change in any way.
- In Moz 1.0 the pages would half load intermittently. Usually, I was able to reload the page a few times to get data. Because Moz re-submits (lame, lame, lame) the page request when you view source I couldn't tell if it was displaying improperly on the client side or not getting the full page from the server side.
- wget & lynx would get the pages once, just fine. However both complained of "disconnected" or "connection reset by peer".
If you really want to get this running using inetd, here's my additional notes on this:
Configure inetd: (used this as reference: http://www.adamswann.com/library/2001/Cygwin-Inetd.html)
from within cygwin shell (zsh) run:
cygrunsrv -I inetd -d "CYGWIN inetd" -p /usr/sbin/inetd -a -d -e CYGWIN=ntsec
add line to c:\%WINDIR%\system32\drivers\etc\services:
cvstrac ####/tcp # CVSTrac server
add line to d:\services\cygwin\etc\inetd.conf:
cvstrac stream tcp nowait root /usr/sbin/cvstrac cvstrac http /home/cvs Project-name
NOTE: if I used any other user (myself or cvs), it wouldn't work via inetd. I'd get "cvstrac: couldn't change directory to /home/cvs" errors
- cvstrac-2.0.1-MinGW-20071107.7z 252796 bytes added by ono on 2007-Nov-07 12:54:12 UTC.
CVSTrac for Windows (MinGW
gcc 4.2.1) 2007-Nov-07 CVS build