match any <
It was motivated by `ddclient`:
* `return <<`:
https://github.com/ddclient/ddclient/blob/e0611ab192d6b8d3d2b09d6868d063ca5591ad02/ddclient.in#L7200-L7223
* `fun(<<)` (although this particular example is not handled by this PR due to
string literal preceding `<<`):
https://github.com/ddclient/ddc
What I'd like to achieve is to access `%buildsubdir` in `%prep` after `%setup`
was called. It appears that RPM first parses and evaluates all macros in
`%prep` and only then executes `%setup` which pushes `%buildsubdir`. So my
question is there any way to defer macro expansion or is there some d
When db is opened rpm registers `rpmlog()` based callback within sqlite and
passes `rpmdb` as user data. When db is closed and `rpmdb` gets eventually
freed it might lead to UAF if another sqlite3 db is opened different than rpm
db.
It happens ie if normal user invokes `dnf download`. First rpm
Completely agree with everything you've said. I've actually had a look at
sqlite API myself before creating PR and couldn't find anything that would work
per database. I didn't feel qualified to make decision on stopping/altering
logging within rpm either. Hence this fix.
I'd say it's not only
**Describe the bug**
Architecture applicability check is not performed early in the process by
`rpmbuild` but very late after successful `%install`.
**To Reproduce**
Steps to reproduce the behavior:
1. Sample spec file:
```
Summary:test
Name: test
Version:1
Release:
jpalus left a comment (rpm-software-management/rpm#3594)
Actually mentioned commit was part of 4.19.0 so that's the first version with
regression but I jumped directly from 4.18.x to 4.20.x hence the impression.
--
Reply to this email directly or view it on GitHub:
https://github.com/rpm-softwa
jpalus created an issue (rpm-software-management/rpm#3594)
**Describe the bug**
rpm 4.20.x fails with ie:
```
warning: group http does not exist - using root
```
even though group exists:
```
$ grep http /etc/group
http::51:nginx
```
**To Reproduce**
Steps to reproduce the behavior:
1. Create a g
jpalus left a comment (rpm-software-management/rpm#3594)
One more observation regarding `lookup_field()`:
https://github.com/rpm-software-management/rpm/blob/b043604b9eb684dc761aeacf55a784632ca0ebcd/lib/rpmug.cc#L56-L92
When matching record is found we don't `break` the `while` loop iterating ove
