Add -f to %_default_patch_flags #1513
Conversation
|
Looking back at patch(1) manual reminds me of why this was left lingering: both --batch and --force take some funny assumptions that are not individually settable, and also not quite what you'd want for rpm use: we wouldn't want rpm to skip anything, we'd like it to error out on those questionable things, but no such option exists. Re-reading up on all that a couple of times, and checking out what exactly is this |
|
Oh and while the commit message gets updated: the patch itself documents what options gets added where, repeating that it the summary is not adding any information. The summary should state what is achieved by that (ie ensure non-interactive patch application), the details and rationale behind this particular flag explained in the message. It may seem obvious and redundant now but a few years down the road, when the next person comes along to suggest something else... |
Builds should always run unattended and never require user interactions. We also don't want unser input to affect the result. Without any options patch will ask for files to patch or whether it should reverse pachtes if they could be applied then. As -t aka --batch does reverse patches and skips patches with wrong version for the Prereq: line it is not suited for rpmbuild. We rather want the build to fail than silently drop (possibly security relevant) patches. We would prefer patch to error out when ever the patch looks not quite matching but -f aka --force is the option available closest to that. See man patch for details Resolves: rpm-software-management#978
|
OK, changed summary and extended the commit message. |
|
Thanks, I like it a lot better this way :) |
This prevents the patch utility to ask questions during builds.
See man patch for details
Resolves: #978