The entry for boinc in portage is dated, so I created an ebuild for their latest stable as of 03 December, 2005. The header for the ebuild needs to be changed, and in order to use boinc_gui, it has to be executed when the console is in /var/lib/boinc, and it has to be executed as root, so could be some security issues there. But, the main thing is that the program works, the only thing I havn't been able to do so far is create a climateprediction.net account, but I'm guessing that has more to do with climateprediction.net than the program. Reproducible: Always Steps to Reproduce: 1. 2. 3.
Created attachment 74013 [details] ebuild for 23 November 2005 snapshot of boinc
*** Bug 114390 has been marked as a duplicate of this bug. ***
The boinc ebuild configure tells me that it needs net-misc/curl (7.15.0 I gess). Adding it to the depend list solves the problem.
Ebuilds should be attached as text/plain and be diffs to the latest version available in portage. This is something I was going to take a look at and just haven't had the time yet so I would be interested in taking a look once you have done this. boinc should be run as its own user, as is the case with the current ebuild, rather than root.
Created attachment 74086 [details] boinc 5.2.14 ebuild (cvs) Due to the amount of new versions of the boinc client - they have jumped from 5.2.4 to 5.2.14 in six weeks - I have created a boinc cvs-ebuild. You can build: -libraries only (-server -client) -client only (-server) -server only (-client) The "X" useflag is a leftover from the 4.* ebuild, the configure process of boinc tries to deduce itself if to use X or not (depending if X is present). I do not know of a way to disable X explicitly, even if X is present. If X is present on the system it always builds the graphical client. The server part of my ebuild is not polished off, and you'll only need to build the server part if you're planning to create your own boinc project (which is what I'm doing at the moment). The server part basically copies everything to /usr/share/BOINC, so you have everything you need on your normal system (not only in the gentoo build directory/sandbox). Dependencies: I have been able to build and run boinc-5.2.x with openssl-0.9.7e contrary to the requirements stated on the website (see: http://boinc.berkeley.edu/build.php) I have been able to run the boinc client as a normal user, so running as root definetly isn't required.
Created attachment 74088 [details] patch for boinc-5.2.13a / 5.2.14 patch needed for the 5.2.14 cvs ebuild
Thanks for taking the time to submit a proposal. I had a quick look over it, and I do not want to make it into a live CVS build. I had already uploaded a version I pulled from CVS tagged as 5.2.14, and this version builds here after applying various patches I found in the Debian repo. It is in package.mask until it receives more testing, as I said diffs are preferable as they let me see the changes you are proposing quickly. I will take another look through your ebuild for any improvements that might be added to the ebuild in portage. It still seems pretty fragile, but I do see the need to update. I also built it with openssl 0.9.7 successfully so I don't know why they are incorrectly stating that dep on their site.
I'll create a diff tomorrow so you have a better overview. I guess the diff will be quite radical, as I've changed quite a few things. The cvs build for me was a solution to the problem that boinc doesn't offer source tarballs. So updating to the next version only involved changing the tag-string in the ebuild.
There are a number of reasons why we prepare tarballs rather than using CVS. The tarball on our mirrors is prepared from the sources tagged 5.2.14 anyway. I really don't understand why you have made some of the dependency changes - we aren't going to throw away previous work without reason. I am open to discussion on different approaches, but the current ebuild has already had a significant amount of work put into it and quite a few issues have already been worked through. That is why we work with diffs. It is a busy week anyway, but I will have more time on the weekend.
I wasn't suggesting changing to the cvs method. Sorry if I gave you that impression. CVS was only meant to ease the problems I had in frequently updating boinc-5.2 versions on many machines. When boinc-5.2 is in portage I'll probably stop using my ebuild as I prefer "upstream" maintenance. I can spare some time at the weekend so I can explain what my ideas were. I'd like to do this on a better medium than in this bug report however (I feel like a bug spammer already). Feel free to contact me on my e-mail.
Created attachment 74243 [details] boinc-5.2.20051123.ebuild.diff Here is a .diff file, not sure if I created it properly, I ran diff boinc-4.72.20050813-r3.ebuild boinc-5.2.20051123.ebuild > boinc-5.2.20051123.ebuild.diff
(In reply to comment #11) > Created an attachment (id=74243) [edit] > boinc-5.2.20051123.ebuild.diff > > Here is a .diff file, not sure if I created it properly, I ran > diff boinc-4.72.20050813-r3.ebuild boinc-5.2.20051123.ebuild > > boinc-5.2.20051123.ebuild.diff Not to put you off, but the stuff you proposed removing in the attachment is all still needed. See my updated ebuild in package.mask - it does require more testing. I am marking as FIXED as I have committed an updated ebuild. Stephen feel free to email me about any ideas if you wish.
