New ebuild for aolserver 4.0-beta2. This ebuild can be easily maintained as aolserver moves to 4.0 final. Reproducible: Always Steps to Reproduce: 1. 2. 3.
Created attachment 9186 [details] aolserver-4.0.ebuild (really for aolserver-4.0 beta2 Uploading first ebuild version. This is aolserver-4.0 beta2, but I have not encoded the beta2 status in the ebuild name (feel free to correct me in a convenient way). This ebuild will be easy to maintain and upgrade as aolserver moves to 4.0 final very soon. Aolserver is a high-performance gpl'd web server backed by AOL and open source since the late '90s. This ebuild should go in net-www/aolserver/ . --Chris
Created attachment 9686 [details] aolserver-4.0 beta3 ebuild. supercedes the previous ebuild. aolserver-4.0 beta3 finishes adding the openacs i18n tcl code and fixes some bugs, including a wait to shutdown bug.
Created attachment 12861 [details] aolserver-4.0_beta8.ebuild (finally got the file name correct even!) Beta bump. search for upcoming associated ebuilds that work with this beta8: nssha1 nsxml nspostgres nscache
Created attachment 12970 [details] aolserver-4.0_beta8.ebuild (updated to copy headers for use in openfts ebuild) I updated the ebuild to copy a few header files which are used in my openfts driver ebuild (search for that).
emerge aolserver (aolserver-4.0_beta8) configure: error: must build tcl with --enable-threads Since Bug 2024 --enable-threads has been disabled in tcl. I enabled it, emerged tcl and still received the same error message.
Yes, it is correct that I had to hack the --enable-threads into the tcl ebuild; but i used the newer tcl-8.4.2-r1 ebuild. When I built that version of tcl enabling threads, then I could proceed with the aolserver ebuild without error. My apologies for not warning of this in the aolserver ebuild; I disagree with the resolution of bug 2024 by disabling threading... But I understand that tcl has been having problems with it. I had forgotten that I had to enable threading in tcl to get my aolserver to build... for those that know ebuilds better than I, what would be a good way to resolve this issue in my ebuild?
I'm having the same problem with --enable-threads. However, it seems that the problem is in emerging Tcl, as it gives an error, /usr/sbin/ebuild.sh: line 36: --enable-threads: command not found This is right after the configuration and right before the compiling begins. Chris, how did you enable threads when you built tcl? talli
You get the command not found because you uncomment the --enable-threads (which is after the || die part, so it doesn't go into the configure command). I had the exact same problem... But it should look like this: ./configure --host=${CHOST} \ --prefix=/usr \ --mandir=/usr/share/man \ --enable-threads \ || die # threading is not recommended as it breaks some packages # --enable-threads \ Instead of this ./configure --host=${CHOST} \ --prefix=/usr \ --mandir=/usr/share/man \ || die # threading is not recommended as it breaks some packages --enable-threads \ Hope that helps! Then your tcl should build, then your aolserver...
Created attachment 19446 [details] aolserver-4.0_beta10.ebuild (with headers needed for use in openfts ebuild) version bump. wonder when this ebuild will make it into cvs? Hopefully soon?
Please see bug #2024 for developments on the TCL threads issue.
Created attachment 19635 [details] aolserver-4.0_beta10.ebuild and associated files I modified chris's ebuild to install aolserver into the standard gentoo filesystem, added service control scripts, a dependency on tcl and a check that tcl was compiled using USE=threads (which you can do using the ebuild attached to bug 2024). Hope that's useful to folks.
Created attachment 19723 [details] aolserver4.0 beta 10 (with module support)
Comment on attachment 19723 [details] aolserver4.0 beta 10 (with module support) Same as lees patch but makes makefile.global available for module compilation
Created attachment 21582 [details] Aolserver 4.0 official release ebuild
To better fit in with Gentoo's way of handling vhosts (/var/www/<hostname>) I've had to make some changes to the ebuild and the default config. Whats different? * config.tcl is installed to /etc/aolserver * serverroot is now /var/aolserver * default pageroot is /var/www/localhost/aolserver/www * each vhost will then be /var/www/<host>/aolserver/www * I've modified config.tcl to use the new pageroot for vhosts, though I have no means of testing it. * sample-config.tcl and the Makefiles go into /usr/share/aolserver * Module configs go into /etc/aolserver/modules/<name>/. Anyone know how to tell nsperm thats where its configs are stored? * Some smaller changes Lots of testing is appreciated, especially the vhost setup! http://dev.gentoo.org/~port001/aolserver/
Created attachment 42174 [details] Aolserver 4.0.8 ebuild I've added the Aolserver 4.0.8 ebuild based on the 4.0.5 version by Ian. It's identical except for some fixes in the init script and I changed the default config file name to aolserver.tcl (to make the init script work).
Created attachment 43293 [details] Aolserver 4.0.8 revised ebuild This should obsolete the 'Aolserver 4.0.8 ebuild' submitted 2004-10-19 09:00 PST
I made some changes to the ebulid and default config.tcl to better match the directory structure conventions of Gentoo, and to generally make the ebuild (in my view) more readable. What's different? * logs are now placed in /var/log/aolserver * the pid file is now placed in /var/run/aolserver * the default pageroot is now /var/www/ns-localhost * each vhost will then be /var/www/ns-<vhost> -or- /var/www/<vhost> - I find this to be a better solution than crafting a double hierarchy by creating an aolserver/www directory below each apache host (/var/www/myhost/aolserver/www
I made some changes to the ebulid and default config.tcl to better match the directory structure conventions of Gentoo, and to generally make the ebuild (in my view) more readable. What's different? * logs are now placed in /var/log/aolserver * the pid file is now placed in /var/run/aolserver * the default pageroot is now /var/www/ns-localhost * each vhost will then be /var/www/ns-<vhost> -or- /var/www/<vhost> - I find this to be a better solution than crafting a double hierarchy by creating an aolserver/www directory below each apache host (/var/www/myhost/aolserver/www besides, who wants to type that?). The way I see it, any content served by AOLserver is likely to be a logically different virtual host anyway, and there is no reason to try to subjugate aolserver content to the apache hierarchy. * set 'enabledtclpages' to true by default in config.tcl * I had some trouble with the mirror://sourceforge method of grabbing AOLserver sources, so I added a set of explicitly specified servers to SRC_URI * Reduced the levels of complexity in how path variables are specified * Spun the code to check that tcl was compiled with threads off to its own function * Made heavier use of the ebuild doins, doexe routines as opposed to cp's * Rearranged support files into directories by version (ie net-www/aolserver/files/4.0.8/config.tcl). I find this makes things easier when it is time to roll an ebuild for a new version. * Various smaller changes.
A couple more notes on 'Aolserver 4.0.8 revised ebuild' (2004-11-04 09:29 PST): * I've tested this ebuild extensively, but additional testing would be appreciated! * The ebuild compiles, installs, merges, and unmerges cleanly. * It seems to me this ebuild is ready to be in CVS? Thanks, -- Travis
Created attachment 43381 [details] Aolserver 4.0.8 revised ebuild (2) Removed the explicitly specified URLs. I thought the better of it. -- Travis
ebuild 4.0.8 won't compile, because pthread_kill_other_threads_np() is missing. I use nptl, which I think makes the function obsolete. Is there a workaround? ebuild 4.0 compiles ok even with nptl.
I've not tried, but it seems pthread_kill_other_threads_np() is now a native feature of the NPTL. Simply commenting it out should be enough. These ebuilds will go into portage as soon I return to develoment (early December time).
Well, its in at last. The modules will have to wait a few more days as I need to evaluate the need for an eclass, and if so, implement it. Leaving this bug open for a few more days for any feedback/problems.
Resolving.
