elinks uses RAND_egd, which is removed from LibreSSL. This patch removes the use of RAND_egd. I have also sent this patch to mainstream. Reproducible: Always
Created attachment 417234 [details, diff] elinks-0.12_pre6-r2.ebuild.patch
Created attachment 417236 [details, diff] elinks-0.12_pre6-libressl.patch
Created attachment 417266 [details] elinks-0.12_pre6-libressl.patch
The LibreSSL devs have put up a page with patches to packages that haven't yet reached compatibility with LibreSSL at http://www.libressl.org/patches.html ; this may be handy for those who are dealing with such packages. :-) They are also sending their patches to upstream as well. Whether or not upstream accepts their patches, well... :-P I adapted the patches on that page for elinks-0.12_pre6-r1; they didn't apply cleanly as they were (I will attach them to this bug). Other than that, the ebuild patch Marek wrote is good (I only had to change the epatch lines to point to the patches I got from the aforementioned page and remove Marek's patch). I think the LibreSSL patch is better because it's less of a "scorched earth" approach than Marek's. No offense is meant Marek; I'm glad you opened this bug and took the time to patch the ebuild. :-)
Created attachment 417360 [details, diff] LibreSSL patch #1
Created attachment 417362 [details, diff] LibreSSL patch #2
As hasufell wrote here https://github.com/heimdal/heimdal/commit/427a60057cbb27cde5f7cf412dfebc52ba709c86 "The EGD daemon is completely unmaintained and has not seen a release since 13 years which is not an acceptable timeframe for cryptographic software. It is not packaged in any linux distribution I know of and definitely not in *BSD." As I see it, the new versions of such programs probably will not work at all with such old systems that have need for EGD daemons, and as it is cryptographicaly insecure to use EGD, programs should not be using them at all. But of course it does not matter to me what patch will be used, as long as Gentoo's support for LibreSSL grows :)
I have no idea why the BSD guys wrote their patch that way then. Also, 100% agreed, I don't care which patch gets used as long as it helps LibreSSL adoption. :-)
Upstream has recently applied half of the libressl patches: http://repo.or.cz/elinks.git/commit/f4a58ba3b574a478fd5954ba2c5b29e8b809ff9b I guess the code between the ifdefs is now dead...
*** Bug 570592 has been marked as a duplicate of this bug. ***
Folks, if you don't CC maintainers in, there's very little chance to have the patch merged into Portage (close to none).
Call me crazy, but I assumed that the system did that for you automagically.
Any news about merging the patches?
(In reply to timofonic from comment #13) > Any news about merging the patches? can you pleaes test elinks-9999.ebuild. following on felix's observation, it should work.
(In reply to Anthony Basile from comment #14) > (In reply to timofonic from comment #13) > > Any news about merging the patches? > > can you pleaes test elinks-9999.ebuild. following on felix's observation, > it should work. actually i just tested. i'm marking this resolved upstream, please reopen if there's still an issue.
