Version 29.0 of MythTV was released on 2017-07-31. (They dropped the leading "0." from the version numbers with this release.) I have an ebuild that at least compiles and runs the new version that I will attach. Reproducible: Always
Created attachment 516856 [details] Ebuild that at least compiles I made changes to the ebuild based on release notes and removed mythlogserver. The resulting program runs, but it won't connect to my backend (possibly due to a version mismatch). The inability to connect to the backend server concerns me. I'm also not an ebuild expert, so the build may be a bit of a hack; I would appreciate any constructive feedback.
From my understanding and experience, MythTV client and server always need to be upgraded in lock-step. A client of one version simply will not talk to a server of another. What's worse, is that when you upgrade the server, it auto-upgrades the database, and there is no way of going back, other than to use a backup from before the upgrade. For a situation like this, I'd suggest making tarballs of /etc/mysql, /var/lib/mysql, /etc/mythtv, and /var/lib/mythconverg. Then upgrade the server and client, upgrade the database, the restart it all and test. The tarballs will make it easier to get back to where you were, if needed. (as opposed to restoring from a database backup)
Created attachment 517306 [details, diff] remove "media-libs/libdvb" from mythtv-29.0.ebuild media-libs/libdvb is no longer in portage.
After removing media-libs/libdvb from the ebuild (patch attached) I was able to compile mythtv-29.0. I used mythtv-setup to upgrade my database. On my system mythfronend-29.0 runs with mythbackend-29.0 -- so for me your ebuild works great!
*** Bug 648214 has been marked as a duplicate of this bug. ***
*** Bug 648564 has been marked as a duplicate of this bug. ***
Preston wanted to proxy maintain this in the duplicated bug report
Yes, I do. I'm still trying to find out how to get started as a proxy maintainer. I think we should move forward with the 0.28.2 ebuild first (as in bug 648564), as that's a minor update, and then go for 29.1.
Any chance of updating this to 29.1 which is released?
Created attachment 525636 [details] mythtv-29.1.ebuild EAPI 6 ebuild
I'm attaching an ebuild for MythTV 29.1 that I just put together. It does in fact compile. This actually uses fixes/29 as of 2018-03-24 as noted below. In order to expressly use the older version that was tagged as 29.1 you can change the BACKPORTS to 9b7b96283458073007a8f3801d5284fb59af3979: https://github.com/MythTV/mythtv/commit/9b7b96283458073007a8f3801d5284fb59af3979 Also note that it was based on the existing EAPI 6 mythtv-0.28.1-r4.ebuild. The notable changes to that included: 1. Obviously the BACKPORTS, where I used 60e40b352ab95a135ec2ab8f9f1ee93b4f9d245e which was the last fixes/29 commit as of 2018-03-24. 2. The change of the MYTHTV_BRANCH to use "fixes/29". 3. The removal of the application of the ${P}-glibc225.patch which is already in 29. 4. The removal of any references to libmp3lame and libfaac which are no longer used (as per the 29.0 release notes). 5. The addition of config changed to enable/disable the new systemd_journal based on the systemd USE flag (just as was already done for systemd_notify). I'll be installing this on my systems soon, which are x86 without systemd, and will report back as to how that goes. Many thanks to Pacho Ramos for the EAPI 6 version 0.28.1 ebuild I based this on!
I upgraded today with this and everything seems 100%. Note that the ebuild I used was actually slightly customized. Specifically mine has the qtdbus requirement removed as well as the requirement for udisks2 when the dvd USE flag is set.
Created attachment 526186 [details] 0.29.1 failed compile @Tom Dexter, I tried to compile your ebuild but it failed. The log is attached.
Hi Rajil. It took me a while to figure out what happened there: You renamed the ebuild to use 0.29.1, which is NOT the correct version. MythTV moved to versions like 29.1 rather than 0.29.1 after 0.28.x. A lot of what goes on in the ebuild is based on the file name and yours is incorrect. Use the original name and that should work. Also note that my ebuild I posted can actually have the line requiring lame removed: >=media-sound/lame-3.93.1 ...and mythtv no longer needs it.
Created attachment 526202 [details] media-plugins/mythplugins-29.1 ebuild (untested) I'm attaching an untested 29.1 version of the mythplugins ebuild. I don't use any of these and have not tested this. Note that just as with the attached mythtv ebuild, you can change the BACKPORTS to 9b7b96283458073007a8f3801d5284fb59af3979 if you want the original 29.1 version.
Created attachment 526216 [details] Failed build @Tom, I renamed the file to mythtv-29.1.ebuild. Unfortunately, the compile failed (log attached). I commented out 'use systemd || myconf="${myconf} $(use_enable mythlogserver)"' to disable mythlogserver and was able to compile successfully. Btw i am using OpenRC.
I'm using OpenRC as well. Your problem is that you have the mythlogserver USE flag set. I think it's always been on by default in the ebuilds, though I'm not sure why. The previous mythtv-0.28.1-r4.ebuild was just like that. I have -mythlogserver in my USE and it works the way it is.
Created attachment 526442 [details] mythtv-29.1.ebuild EAPI 6 ebuild with missing python deps I'm attaching an updated version of mythtv-29.1.ebuild which includes dependencies for dev-python/future and dev-python/requests-cache, which appear to be required for the ttvdb.py grabber.
One thing to note (and this change is NOT in the attached 29.1 ebuild): I believe that somewhere between the ebuilds for 0.28 and 0.28.1 the Qt dependencies in the ebuild were changes so as to not trigger a rebuild of mythtv for minor slot upgrades to Qt...for example from 5.7 to 5.9. The entries in COMMON where changed from, for example: dev-qt/qtcore:5= to dev-qt/qtcore:5. Based on this: https://lists.gt.net/mythtv/users/616874 I've changed my own ebuild so as to define those in COMMON as: dev-qt/qtcore:5= dev-qt/qtgui:5= dev-qt/qtnetwork:5= dev-qt/qtscript:5= dev-qt/qtsql:5=[mysql] dev-qt/qtopengl:5= dev-qt/qtwebkit:5= dev-qt/qtwidgets:5= dev-qt/qtxml:5= ...though it sounds as though it may not actually be necessary. If at some point an ebuild is added to portage, one change I would suggest would be to change the IUSE to have "mythlogserver" rather than "+mythlogserver". That is, to remove the default on status for that when systemd is not in use. I believe that's a holdover from when mythlogserver was actually required. I believe it's now generally only used by developers.
Due to bug 656250 the DEPEND needs changing to use x11-base/xorg-proto
Hi i can confirm that all the issues and its solutions here. 1. I ran into 0.29.1 Ebuild error. Solved by 29.1 ebuild Name. 2. I ran into +mythlogserver. Solved by disable the Useflag. Notes: I run the Server mythbackend on OpenRC and install the client mythfrontend on Systemd. So on the Systemd i had no issues to compile it with +mythlogserver, because the ./configure comand did not set --enable-mythlogserver. I think it would fail too if its set. On OpenRC, it faild with like the attachment 526216 [details] https://www.mythtv.org/wiki/Mythlogserver The Version without +mythlogserver creates logs on my Backend OpenRC as usual in /var/log/mythtv just the process mythlogserver is not there. The Ebuild i used was Tom Dexter attachment 526442 [details] with his qt in COMMON Changes. Thank you all for your work in that ebuild!
You're quite welcome! I hadn't gotten any emails on any of the recent changes here, as apparently I never added myself to the cc list. Yea, as I noted above the IUSE in that should probably have +mythlogserver changed to just mythlogserver to remove that default. I'll probably be adding a new ebuild here in the near future for the current 29/fixes. That will include the ebuild fix to use x11-base/xorg-proto as noted by Al Johnson (thanks!). It will also include the addition of the dependency for virtual/pkgconfig which is related to bug 658534 and the freetype changes. The rest of that issue is addressed in the current 29/fixes. I was a little concerned as to bug 649044 regarding the compile failure of mythcdrom-linux.cpp but looking at the patch attached to that one, that appears to be fixed in the current code as well and possibly even in the original 29.1. I'll probably also change all the qt dependencies to use just :5 rather than :5= as I really don't think there's any need to require a recompile for minor qt version changes.
One important thing I ran into when creating this new ebuild: My ebuild like the existing 0.28.1-r4 ebuild in portage used these: SRC_URI="https://github.com/MythTV/mythtv/archive/${BACKPORTS}.tar.gz -> ${P}.tar.gz" S="${WORKDIR}/${P}/mythtv" One very ugly aspect of that was that a subsequent addition or change in the revision (-r1 for example) would reuse the existing distfile archive as well as it's hash and NOT the desired git version at all. I was using 29.1-r1 for my new ebuild, and to prevent that issue, I changed both of the above lines to use ${PF} in place of ${P}. That is, with the -r1 in the file name and source path. I figured I'd point that out in case someone else was attempting to make these changes on their own.
Created attachment 544192 [details] New mythtv-29.1-r1.ebuild 29.1 ebuild for fixes/29 I've attached a mythtv-29.1-r1.ebuild ebuild that should have all the recent fixes and installs the current fixes/29 (as of a commit on 8/14/2018 specifically). Also note that I've changed it do as to not default to having mythlogserver enabled. If anyone else is interested in mythtv without having dbus installed, as per my discussion on bug 580856, here's a diff of the attached ebuild with the one I use: --- mythtv-29.1-r1.ebuild-STOCK 2018-08-20 12:41:20.000000000 -0400 +++ mythtv-29.1-r1.ebuild 2018-08-20 10:26:33.000000000 -0400 @@ -32,7 +32,6 @@ COMMON=" dev-libs/glib:2 dev-qt/qtcore:5 - dev-qt/qtdbus:5 dev-qt/qtgui:5 dev-qt/qtnetwork:5 dev-qt/qtscript:5 @@ -67,7 +66,6 @@ ) dvd? ( dev-libs/libcdio:= - sys-fs/udisks:2 ) egl? ( media-libs/mesa[egl] ) fftw? ( sci-libs/fftw:3.0= ) If you have your dvd drive path expressly configured in MythTV and don't need udisks2 for drive detection, this works fine. Note however that I don't have any bluray drives, but I'd guess that removing it's udisks2 dependency would work as well.
The bug has been closed via the following commit(s): https://gitweb.gentoo.org/repo/gentoo.git/commit/?id=a9f434f613b2c031adcc77b569baf05db6ef02af commit a9f434f613b2c031adcc77b569baf05db6ef02af Author: Pacho Ramos <pacho@gentoo.org> AuthorDate: 2018-10-20 14:47:11 +0000 Commit: Pacho Ramos <pacho@gentoo.org> CommitDate: 2018-10-20 16:35:30 +0000 media-tv/mythtv: Bump to 29.1 Thanks-to: Preston Crow Thanks-to: Tom Dexter Closes: https://bugs.gentoo.org/645850 Closes: https://bugs.gentoo.org/649044 Closes: https://bugs.gentoo.org/658534 Closes: https://bugs.gentoo.org/665182 Signed-off-by: Pacho Ramos <pacho@gentoo.org> Package-Manager: Portage-2.3.51, Repoman-2.3.11 media-tv/mythtv/Manifest | 1 + media-tv/mythtv/mythtv-29.1.ebuild | 369 +++++++++++++++++++++++++++++++++++++ 2 files changed, 370 insertions(+)