[master][makeupdates][PATCH 0/3] Makeupdates version handling improvements
by Martin Kolman
The first patches add support for fetching older packages for a newer
Fedora release. This is 100% valid: while the package might have been built
for an older release, provided there was no update since then it is still
valid (and tagged in Koji) for newer Fedora releases.
The second patch makes sure such packages are picked up from the package cache.
They are stored in a separate folder per Fedora release, so that makeupdates
can discern between packages that are really outdated and those that were just not
updated in the latest Fedora release.
The third patch fixes issues with long build ids in some packages,
that were preventing those packages to be picked-up from the cache.
Martin Kolman (3):
Fetch older valid releases
Store older valid packages in separate folder
Fix handling of long release ids
scripts/makeupdates | 103 ++++++++++++++++++++++++++++++++++++++++++----------
1 file changed, 83 insertions(+), 20 deletions(-)
--
1.8.4.2
10 years, 5 months
[master/f20-branch][PATCH] Don't try to investigate empty string for unicode chars (#1035799)
by Vratislav Podzimek
If no size is given, we shouldn't try to investigate the string. Treat it as an
english spec instead which means falling back to the old behaviour.
Signed-off-by: Vratislav Podzimek <vpodzime(a)redhat.com>
---
pyanaconda/ui/gui/spokes/custom.py | 2 +-
1 file changed, 1 insertion(+), 1 deletion(-)
diff --git a/pyanaconda/ui/gui/spokes/custom.py b/pyanaconda/ui/gui/spokes/custom.py
index b142364..fcbb859 100644
--- a/pyanaconda/ui/gui/spokes/custom.py
+++ b/pyanaconda/ui/gui/spokes/custom.py
@@ -153,7 +153,7 @@ def size_from_entry(entry):
try:
# if no unit was specified, default to MB. Assume that a string
# ending with any kind of a letter has a unit suffix.
- if unicodedata.category(size_text[-1]).startswith("L"):
+ if size_text and unicodedata.category(size_text[-1]).startswith("L"):
size = Size(spec=size_text)
else:
size = Size(en_spec="%sMB" % size_text)
--
1.8.4.2
10 years, 5 months
[master][PATCH] Import Xkl only when really needed
by Vratislav Podzimek
This prevents a traceback when using only the anaconda-core and anaconda-tui
packages that doesn't require the libxklavier package. There's no point in
having libxklavier and importing Xkl without X server running, it does nothing
and fails to initialize.
Signed-off-by: Vratislav Podzimek <vpodzime(a)redhat.com>
---
pyanaconda/keyboard.py | 4 ++--
1 file changed, 2 insertions(+), 2 deletions(-)
diff --git a/pyanaconda/keyboard.py b/pyanaconda/keyboard.py
index 21505a2..0fe7d6e 100644
--- a/pyanaconda/keyboard.py
+++ b/pyanaconda/keyboard.py
@@ -49,7 +49,7 @@ from pyanaconda.safe_dbus import DBUS_SYSTEM_BUS_ADDR, DBusPropertyError
from pyanaconda.constants import DEFAULT_VC_FONT, DEFAULT_KEYBOARD, THREAD_XKL_WRAPPER_INIT
from pyanaconda.threads import threadMgr, AnacondaThread
-from gi.repository import Xkl, Gio, GLib
+from gi.repository import Gio, GLib
import logging
log = logging.getLogger("anaconda")
@@ -393,7 +393,7 @@ class XklWrapper(object):
return XklWrapper._instance
def __init__(self):
- from gi.repository import GdkX11
+ from gi.repository import GdkX11, Xkl
#initialize Xkl-related stuff
display = GdkX11.x11_get_default_xdisplay()
--
1.8.4.2
10 years, 5 months
[initial-setup][master] Adapt to changes in anaconda tui spoke categories (#1035462)
by Vratislav Podzimek
The anaconda commit d4417b299d9732b2c7608f55384395ec4ce30172 renames the
'password' category to the 'user' category. We need to adapt to that change
otherwise User and Root password spokes are omitted in the text mode Initial
Setup.
Signed-off-by: Vratislav Podzimek <vpodzime(a)redhat.com>
---
initial_setup/tui/hubs/initial_setup_hub.py | 2 +-
1 file changed, 1 insertion(+), 1 deletion(-)
diff --git a/initial_setup/tui/hubs/initial_setup_hub.py b/initial_setup/tui/hubs/initial_setup_hub.py
index 3b6387e..093455f 100644
--- a/initial_setup/tui/hubs/initial_setup_hub.py
+++ b/initial_setup/tui/hubs/initial_setup_hub.py
@@ -10,7 +10,7 @@ _ = lambda x: gettext.ldgettext("initial-setup", x)
N_ = lambda x: x
class InitialSetupMainHub(TUIHub):
- categories = ["password", "localization"]
+ categories = ["user", "localization"]
prod_title = product.product_title()
if prod_title:
--
1.8.4.2
10 years, 5 months
[master][rhel7-branch] Some advanced storage spoke patches, mostly s390x-related
by Samantha N. Bueno
First just adds the ability for users to manually add zFCP devices
from the advanced storage spoke. Following this, it should no longer be
necessary to specify a zFCP device in a parm file. Celebrate.
Second breathes life into the lifeless skeleton of a z-panel that once
was. I'm having some residual issues on this one and decided to post it
for review anyway--the issues are minor and maybe another pair of eyes
can help.
Filtering devices by CCW (device id), LUN, and WWPN won't work for some
reason. If I enter a LUN, this should occur:
self._lunEntry = self.builder.get_object("zLUNEntry")
....
self._lunEntry.get_text()
However, the text is always stored in one of the other two entries
(zCCWEntry or zWWPNEntry). This behavior occurs with all three of the
entries. Not sure what's going on there.
Also, the WWPN column in the treeview should be displaying the WWPN
obviously, except....it isn't. Having a bit of trouble figuring out how
to make this show up.
Third patch fixes an issue where searching by LUN in the advanced
storage spoke failed.
Lastly, here's a screencast of the patches in action:
http://sbueno.fedorapeople.org/s390x-gui-patches.webm
If you want to have a poke yourself, I'm glad to give access to my s390x
guest.
Samantha
10 years, 5 months
[master] libtimezonemap updates
by David Shea
I made a few tweaks to the libtimezonemap API to improve the handling of the
Etc zones and incorporated them into the anaconda patch. The behavior now is to
unset the pin and highlight the selected zone. The previous behavior was to
unset the pin and highlight GMT.
You may notice as you select the Etc zones that the highlight seems backwards;
e.g., Etc/GMT+5 selects the Eastern US and Canada and Colombia, in UTC-5. This
is expected and it's because POSIX is freaking weird and defines + as hours
west of Greenwich for some damn reason.
I didn't make any changes regarding not sending a location change signal when
setting the location from a click, but I haven't noticed any problems from
that. If it was from the Etc zones, it probably shouldn't be an issue, since
the location-changed signal emitted by clear_location will have a NULL location
pointer, and no change will be made to the combo boxes. If there was something
else there, well, make your case.
10 years, 5 months
[master/f20-branch][PATCH] Do not write out the vconsole.keymap boot option (#1035316)
by Vratislav Podzimek
It no longer does any good and just causes potential problems with changing
VConsole keymap in the configuration file and regenerating initrd.img with
dracut.
Signed-off-by: Vratislav Podzimek <vpodzime(a)redhat.com>
---
pyanaconda/keyboard.py | 4 ----
1 file changed, 4 deletions(-)
diff --git a/pyanaconda/keyboard.py b/pyanaconda/keyboard.py
index 2bfdb00..99b4f64 100644
--- a/pyanaconda/keyboard.py
+++ b/pyanaconda/keyboard.py
@@ -263,11 +263,7 @@ def dracut_setup_args(keyboard):
"""
- if not keyboard.vc_keymap:
- populate_missing_items(keyboard)
-
args = set()
- args.add("vconsole.keymap=%s" % keyboard.vc_keymap)
args.add("vconsole.font=%s" % DEFAULT_VC_FONT)
return args
--
1.8.4.2
10 years, 5 months
[master][blivet][PATCH] Consider existing None format nonresizable (#1033778)
by Vratislav Podzimek
If there is an existing format, but we cannot recognize it we shouldn't consider
it resizable.
Signed-off-by: Vratislav Podzimek <vpodzime(a)redhat.com>
---
blivet/devices.py | 3 +--
1 file changed, 1 insertion(+), 2 deletions(-)
diff --git a/blivet/devices.py b/blivet/devices.py
index d8ca488..c0cc809 100644
--- a/blivet/devices.py
+++ b/blivet/devices.py
@@ -656,8 +656,7 @@ class StorageDevice(Device):
def resizable(self):
""" Can this type of device be resized? """
return (self._resizable and self.exists and
- (self.format.type is None or self.format.resizable or
- not self.format.exists))
+ (not self.format.exists or self.format.resizable))
def notifyKernel(self):
""" Send a 'change' uevent to the kernel for this device. """
--
1.8.4.2
10 years, 5 months
Re: [fedora-arm] [master/rhel7-branch][PATCH] Don't allow bootloader and /boot on iSCSI on s390 (#1034222)
by Dennis Gilmore
-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1
El Tue, 26 Nov 2013 22:10:51 +0000
Peter Robinson <pbrobinson(a)gmail.com> escribió:
> On Tue, Nov 26, 2013 at 9:39 PM, Dennis Gilmore <dennis(a)ausil.us>
> wrote:
> > -----BEGIN PGP SIGNED MESSAGE-----
> > Hash: SHA1
> >
> > El Tue, 26 Nov 2013 14:17:22 +0000
> > Peter Robinson <pbrobinson(a)gmail.com> escribió:
> >> On Tue, Nov 26, 2013 at 2:01 PM, Dan Horák <dan(a)danny.cz> wrote:
> >> > On Tue, 26 Nov 2013 14:50:29 +0100
> >> > Vratislav Podzimek <vpodzime(a)redhat.com> wrote:
> >> >
> >> >> On Tue, 2013-11-26 at 14:47 +0100, Dan Horák wrote:
> >> >> > On Tue, 26 Nov 2013 14:41:15 +0100
> >> >> > Vratislav Podzimek <vpodzime(a)redhat.com> wrote:
> >> >> >
> >> >> > > s390(x) cannot boot from an iSCSI disk.
> >> >> >
> >> >> > what about other arches eg. ARM?
> >> >> Is there any bug on that? And any HW/person for testing?
> >> >
> >> > I think boot from iSCSI should be enabled only for positive list
> >> > of arches (like x86, ppc), I guess it requires special support in
> >> > device firmware
> >> >
> >> > the Fedora/ARM list is in on CC
> >>
> >> It depends, you can boot using an initrd with SW iSCSI support
> >> (either off a local card or via netboot) and have the rootfs on an
> >> iSCSI LUN and there's a few people have done that with the
> >> Trimslice as a PCIe attached gig interface tends to be faster than
> >> usb2 attached SSD.
> >>
> >> If you mean via an iSCSI HW device with appropriate firmware and
> >> firmware provided iSCSI LUN where the OS just sees the
> >> usual /dev/sdX style devices we don't currently see that
> >> functionality but I wouldn't be surprised if it was available of
> >> some Server SoCs soon.
> >
> > I don't know of any u-boot supoort for iscsi, but its always
> > possible. I expect that aarch64 from day 1 will support iSCSI. I
> > think that the 32 bit devices that have UEFI should have support
> > also.
>
> I wasn't referring to iSCSI via uboot, there's other means of
> achieving it.
Right and that was clear in your email, I was filling in some missing
details from your email. as there is multiple ways for an iscsi target
to be enumerated and your answer only partially answered the question.
Dennis
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v2.0.22 (GNU/Linux)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=2QB7
-----END PGP SIGNATURE-----
10 years, 5 months
[master] Use format's name instead of type
by Vratislav Podzimek
These anaconda and blivet patches should improve the information about formats
presented to users in the resize dialog. They are inspired by the bug #1033778
that would, however, need some additional changes to be fixed.
Vratislav Podzimek (1):
Use format names instead of types in the resize dialog
pyanaconda/ui/gui/spokes/lib/resize.py | 4 ++--
1 file changed, 2 insertions(+), 2 deletions(-)
--
1.8.4.2
10 years, 5 months