View previous topic :: View next topic |
Author |
Message |
genterminl Guru
Joined: 12 Feb 2005 Posts: 523 Location: Connecticut, USA
|
Posted: Mon Jul 11, 2016 7:39 pm Post subject: apparent inconsistencies kde-apps 16.04.3 fetch restrictions |
|
|
This is the first time I have seen the big bold fetch restriction message in response to an emerge --update. However, an emerge world currently gives me six kde-apps 16.04.3 packages, but only three of them have the fetch restriction attached. I finally figured out that the three packages in kde-apps in the main tree do not have the restriction, but the three in the kde overlay do have the restriction. It appears to be done in the kde5.eclass in the overlay but not in the main tree. I'm curious if this is intentional or just an oversight. (These are all slot :5 apps, in case it matters.) I also have not yet managed to dig out the official upstream release date for these apps, but I'm sure it is soon but still in the future.
Also - I'm curious why some apps (within the same 16.04.3 release cycle) are in the main tree and others are only in the overlay. |
|
Back to top |
|
|
asturm Developer
Joined: 05 Apr 2007 Posts: 8935
|
Posted: Mon Jul 11, 2016 8:11 pm Post subject: |
|
|
16.04.3 release is tomorrow, there can't be ebuilds in tree yet.
Code: | gentoo.git $ git pull
gentoo.git $ find kde-apps/ -name "*16.04.3*ebuild"
gentoo.git $ |
...no result. |
|
Back to top |
|
|
genterminl Guru
Joined: 12 Feb 2005 Posts: 523 Location: Connecticut, USA
|
Posted: Mon Jul 11, 2016 8:55 pm Post subject: |
|
|
Ah, I was partly mistaken The overlay has 279 files kde-apps/xxxxx-16.04.3.ebuild. The main tree has none.
emerge -auDvN world: | Calculating dependencies... done!
[ebuild U #] kde-apps/kholidays-16.04.3:5::kde [16.04.2:5::kde] USE="-debug {-test}" 172 KiB
[ebuild FU #] kde-apps/kcontacts-16.04.3:5::kde [16.04.2:5::kde] USE="-debug {-test}" 133 KiB
[ebuild U #] kde-apps/kpimtextedit-16.04.3:5::kde [16.04.2:5::kde] USE="-debug -designer (-speech) {-test}" 77 KiB
[ebuild FU #] kde-apps/kidentitymanagement-16.04.3:5::kde [16.04.2:5::kde] USE="-debug {-test}" 36 KiB
[ebuild FU #] kde-apps/kontactinterface-16.04.3:5::kde [16.04.2:5::kde] USE="-debug" 25 KiB | Most of these packages, all available ebuilds are in the overlay
eix kcontacts: | [U] kde-apps/kcontacts [1]
Available versions: (5) {M}(~)16.04.2 {M}(~)16.04.3^f **16.04.49.9999 **9999
{debug test}
Installed versions: 16.04.2(5)(05:53:16 PM 06/20/2016)(-debug -test)
Homepage: https://www.kde.org/
Description: Address book API based on KDE Frameworks
[1] "kde" /var/lib/layman/kde | only one has an older ebuild in the main tree.
eix kholidays: | [U] kde-apps/kholidays
Available versions: (5) (~)16.04.2^t {M}(~)16.04.3^t[1] **16.04.49.9999^t[1] **9999^t[1]
{debug test}
Installed versions: 16.04.2(5)^t[1](06:44:42 PM 06/20/2016)(-debug -test)
Homepage: https://www.kde.org/
Description: Library to determine holidays and other special events for a geographical region
[1] "kde" /var/lib/layman/kde | So, I'm still not sure why not all of them have the fetch restriction. There is then the separate question of why the ebuilds were released a bit early. At least they were (mostly?) marked with an appropriate fetch restriction, so you couldn't build them anyway, if you didn't know where to find them. |
|
Back to top |
|
|
asturm Developer
Joined: 05 Apr 2007 Posts: 8935
|
Posted: Mon Jul 11, 2016 9:58 pm Post subject: |
|
|
The fetch restriction is done once, for all of KDE Applications 16.04.3, there is no separate marking. So this is odd. |
|
Back to top |
|
|
genterminl Guru
Joined: 12 Feb 2005 Posts: 523 Location: Connecticut, USA
|
Posted: Mon Jul 11, 2016 10:35 pm Post subject: |
|
|
Yeah, I call it "phase of the moon, and blue magic dust." I just tried emerging kholidays, and it failed to download from anywhere, and then at the end it gave me the fetch restriction message. I tried emerge --update world again, and it didn't show restrictions on kholidays or kpimtextedit, but just trying to emerge kpimtextedit also failed to download, and then at the end gave the fetch restriction message.
It would be nice to track down the cause, but I suppose it will be moot tomorrow, once the release is official. |
|
Back to top |
|
|
genterminl Guru
Joined: 12 Feb 2005 Posts: 523 Location: Connecticut, USA
|
Posted: Tue Jul 12, 2016 4:12 pm Post subject: |
|
|
Any idea what should be the official timing of the release? It's now (if my conversion is correct) 1700 UTC. I just did an eix-sync but still have the fetch restrictions in place. Explicitly emerging kholidays failed download from several mirrors but succeeded from download.kde.org. |
|
Back to top |
|
|
Perfect Gentleman Veteran
Joined: 18 May 2014 Posts: 1248
|
Posted: Tue Jul 12, 2016 4:58 pm Post subject: |
|
|
compiling packages at time of writing this message |
|
Back to top |
|
|
genterminl Guru
Joined: 12 Feb 2005 Posts: 523 Location: Connecticut, USA
|
Posted: Tue Jul 12, 2016 9:03 pm Post subject: |
|
|
Very odd. One last emerge --update still showed the fetch restriction message, but they all updated just fine. |
|
Back to top |
|
|
asturm Developer
Joined: 05 Apr 2007 Posts: 8935
|
Posted: Tue Jul 12, 2016 11:46 pm Post subject: |
|
|
Because you need to sync the overlay. |
|
Back to top |
|
|
kensington Developer
Joined: 02 Jan 2013 Posts: 177 Location: Australia
|
Posted: Wed Jul 13, 2016 2:24 pm Post subject: |
|
|
genstorm wrote: | The fetch restriction is done once, for all of KDE Applications 16.04.3, there is no separate marking. So this is odd. |
RESTRICT in the eclass doesn't stack with the ebuild like IUSE or DEPEND does, so any eg. RESTRICT="test" in the ebuild will override the eclass's fetch restriction. |
|
Back to top |
|
|
genterminl Guru
Joined: 12 Feb 2005 Posts: 523 Location: Connecticut, USA
|
Posted: Wed Jul 13, 2016 3:15 pm Post subject: |
|
|
Kensington: thanks for that explanation - I'm pretty sure all the packages that didn't show the fetch restriction did show a test restriction. Should that be considered a bug in the eclass mechanism (portage?) or at least a "wishlist" to change it? (I haven't checked bgo yet.) In this case, I would think the fetch restriction should override the test restriction, or at least not get overridden by it. |
|
Back to top |
|
|
kensington Developer
Joined: 02 Jan 2013 Posts: 177 Location: Australia
|
Posted: Wed Jul 13, 2016 3:20 pm Post subject: |
|
|
I did previously start a brief discussion about making it automatically stack like the other variables in a future EAPI, but there wasn't much interest.
The test restrict overrides the fetch restriction simply because of how they're declared - RESTRICT="fetch" in the eclass then RESTRICT="test" in the ebuild. We could workaround the issue by switching to RESTRICT+=" test" instead. |
|
Back to top |
|
|
|