On Wed, 2011-02-09 at 17:11 +0200, Panu Matilainen wrote:
> On 02/09/2011 04:55 PM, James Antill wrote:
> > On Wed, 2011-02-09 at 16:44 +0200, Panu Matilainen wrote:
> >> On 02/09/2011 04:40 PM, James Antill wrote:
> >>>Pretty sure we can't remove this, due to API compat. Otherwise, ACK.
> >>
>
On Wed, 2011-02-09 at 17:07 +0200, Panu Matilainen wrote:
> ...and here lies the key to the mystery of yum needing to repopulate
> the transaction after a test-run, doh :)
> ---
> rpmUtils/transaction.py |2 ++
> 1 files changed, 2 insertions(+), 0 deletions(-)
ACK.
On 02/09/2011 04:55 PM, James Antill wrote:
On Wed, 2011-02-09 at 16:44 +0200, Panu Matilainen wrote:
On 02/09/2011 04:40 PM, James Antill wrote:
Pretty sure we can't remove this, due to API compat. Otherwise, ACK.
So instead of removing it, add a comment it's unused?
Yeh, that's fine.
...and here lies the key to the mystery of yum needing to repopulate
the transaction after a test-run, doh :)
---
rpmUtils/transaction.py |2 ++
1 files changed, 2 insertions(+), 0 deletions(-)
diff --git a/rpmUtils/transaction.py b/rpmUtils/transaction.py
index e8f4459..06b5878 100644
--- a/
On Wed, 2011-02-09 at 16:25 +0200, Panu Matilainen wrote:
> In older rpm versions ts.scriptFd is an invisible write-only
> pseudo-attribute hack, so __getattr__() tricks in the transaction wrapper
> wont work for it. Add a new wrapper method for setting it to make it
> less gruesome, accept any fil
On Wed, 2011-02-09 at 16:44 +0200, Panu Matilainen wrote:
> On 02/09/2011 04:40 PM, James Antill wrote:
> > Pretty sure we can't remove this, due to API compat. Otherwise, ACK.
>
> So instead of removing it, add a comment it's unused?
Yeh, that's fine.
On 02/09/2011 04:40 PM, James Antill wrote:
On Wed, 2011-02-09 at 15:59 +0200, Panu Matilainen wrote:
ts.check() is not really an optional call, as there are dependencies
which yum does not check for (notably rpmlib() dependencies). Always
call ts.check(), looking at the return values.
[...]
On Wed, 2011-02-09 at 14:53 +0200, Panu Matilainen wrote:
> Initialize + populate the transaction set to be used for the test-run
> before calling rpm_check_debug to let them use the same set, avoiding
> one round of the fairly expensive re-population of the entire transaction.
> ---
> cli.py
On Wed, 2011-02-09 at 16:11 +0200, Panu Matilainen wrote:
> With newer rpm versions these would just raise AttributeErrors,
> in older versions they are broken to various degrees and yum
> never used them for anything anyway.
> ---
> rpmUtils/transaction.py |4
> 1 files changed, 0 insert
On Wed, 2011-02-09 at 15:59 +0200, Panu Matilainen wrote:
> ts.check() is not really an optional call, as there are dependencies
> which yum does not check for (notably rpmlib() dependencies). Always
> call ts.check(), looking at the return values.
[...]
> --- a/yum/config.py
> +++ b/yum/config.py
In older rpm versions ts.scriptFd is an invisible write-only
pseudo-attribute hack, so __getattr__() tricks in the transaction wrapper
wont work for it. Add a new wrapper method for setting it to make it
less gruesome, accept any file-like object as the fd argument.
---
rpmUtils/transaction.py |
With newer rpm versions these would just raise AttributeErrors,
in older versions they are broken to various degrees and yum
never used them for anything anyway.
---
rpmUtils/transaction.py |4
1 files changed, 0 insertions(+), 4 deletions(-)
diff --git a/rpmUtils/transaction.py b/rpmUti
ts.check() is not really an optional call, as there are dependencies
which yum does not check for (notably rpmlib() dependencies). Always
call ts.check(), looking at the return values.
---
cli.py | 41 -
yum/__init__.py | 37 ++--
Nothing in yum cares for the dependency-tuples-of-doom that the
rpm-level ts.check() returns, having this in transaction wrapper check()
method seems like a more sensible place than in yum base class with
a funny name
(this requires the earlier ts population patch)
---
cli.py |
On 02/09/2011 02:37 PM, Panu Matilainen wrote:
From: Panu Matilainen
Initialize + populate the transaction set to be used for the test-run
before calling rpm_check_debug to let them use the same set, avoiding
one round of the fairly expensive re-population of the entire transaction.
Bah, sorry
From: Panu Matilainen
Initialize + populate the transaction set to be used for the test-run
before calling rpm_check_debug to let them use the same set, avoiding
one round of the fairly expensive re-population of the entire transaction.
---
cli.py | 12 ++--
yum/__init__.py |
Initialize + populate the transaction set to be used for the test-run
before calling rpm_check_debug to let them use the same set, avoiding
one round of the fairly expensive re-population of the entire transaction.
---
cli.py | 12 ++--
yum/__init__.py | 18 ++
On 01/25/2011 05:55 PM, seth vidal wrote:
On Tue, 2011-01-04 at 14:37 +0200, Panu Matilainen wrote:
Hi,
This is something I've been intending to comment on / patch for a long
time but just haven't gotten around...
I know the "debug" part originates from times when yum's own depsolver was
just
18 matches
Mail list logo
