> OK, just hard coding one file attribute doesn't seem like a good idea. I
> added a macro that allows you to register an arbitrary number of local file
> attributes and generators.
Nice, now we can argue if the `__local_file_attrs` is the right name or rather
e.g. `__package_local_file_attrs`
@voxik commented on this pull request.
> @@ -866,6 +866,7 @@ RPMTEST_CHECK([
RPMDB_INIT
runroot rpmbuild -bb --quiet \
+ --define '__local_file_attrs local_generator' \
Maybe something like `totally_random_generator_name`
--
Reply to this email directly or view it on
For someone like me who's utterly unaware of the Ruby ecosystem, setting up the
jekyll server for locally testing docs changes is a rather terrifying
experience.
We should have a cmake target that sets it up and runs it in a hands-free
manner. Something along the lines of "make docs-server"
> I am thinking about actually shipping the dependency generator but also using
> it in the package itself.
This ^^ actually is my use case and here is how it looks in practice:
https://src.fedoraproject.org/rpms/ruby/blob/308b2c0ab2c6268847ed3bf2008e74bce334e810/f/ruby.spec#_187-190
musl doesnt need to use libintl from GNU gettext because it has its own
NLS implementation. Assuming _nl_msg_cat_cntr exists breaks the build against
the musl system.
```
/usr/lib/gcc/x86_64-gentoo-linux-musl/13/../../../../x86_64-gentoo-linux-musl/bin/ld:
../lib/librpm.so.10.0.1: undefined
@voxik commented on this pull request.
> @@ -866,6 +866,7 @@ RPMTEST_CHECK([
RPMDB_INIT
runroot rpmbuild -bb --quiet \
+ --define '__local_file_attrs local_generator' \
While this is "just test case", maybe the `local_generator` could use different
name, just to avoid the
Hmm, I guess we are taking about two different use cases. You are thinking
about a dependency generator that is in the package only and not installed
ever. For that you want a unique name that won't ever clash with installed file
attributes.
I am thinking about actually shipping the dependency
This is a followup to https://github.com/rpm-software-management/rpm/pull/2852/
There is still a number of soft dependencies in
https://github.com/rpm-software-management/rpm/blob/master/CMakeLists.txt:
```
find_package(BZip2)
find_package(Iconv)
...
pkg_check_modules(ZSTD IMPORTED_TARGET
I think the issues is reproducibility and correctness. If you fix the
dependency generator in your package you don't want to old - possibly broken -
deps still in your package, just because the old package is still on your build
system.
--
Reply to this email directly or view it on GitHub:
On the second thought, maybe it really is the right way. I am coming from place
where no "local" generator us was possible, so one was already great
improvement. But maybe there are use cases for multiple generators?
OTOH, one could call them via some helper script or by chaining them (in the
> I think the issues is reproducibility and correctness. If you fix the
> dependency generator in your package you don't want to old - possibly broken
> - deps still in your package, just because the old package is still on your
> build system.
That is actually good point.
So in the Ruby
It seems like the commit hash mentioned in the commit message isn't correct
(it's the "Bump CI" commit which doesn't seem to have anything to do with this).
--
Reply to this email directly or view it on GitHub:
https://github.com/rpm-software-management/rpm/pull/2843#issuecomment-1898909235
You
@dmnks commented on this pull request.
>
return 0;
}
const char * rpmugUname(uid_t uid)
{
-static uid_t lastUid = (uid_t) -1;
-static char * lastUname = NULL;
-
-if (uid == (uid_t) -1) {
- lastUid = (uid_t) -1;
Ah, this is not needed now that we have a `struct`,
There are often a large number of script source files.
While the xargs will later discover that these files are simple text files, it
makes a very large number of io calls to determine this.
Since many people are using containerized builds, the expense of this IO ends
up adding minutes to some
@dmnks commented on this pull request.
>
return 0;
}
const char * rpmugUname(uid_t uid)
{
-static uid_t lastUid = (uid_t) -1;
-static char * lastUname = NULL;
-
-if (uid == (uid_t) -1) {
- lastUid = (uid_t) -1;
I wonder if this line shouldn't be kept?
@dmnks approved this pull request.
Other than the commit note above, the patches look good to me.
--
Reply to this email directly or view it on GitHub:
https://github.com/rpm-software-management/rpm/pull/2843#pullrequestreview-1830246698
You are receiving this because you are subscribed to
Docs still assume that systemd_sysusers is to be used.
--
Reply to this email directly or view it on GitHub:
https://github.com/rpm-software-management/rpm/issues/2857
You are receiving this because you are subscribed to this thread.
Message ID: ___
@ffesti pushed 1 commit.
5059816af9185cc7c3f19ad729b4dc657f912cde Rename __local_file_attrs to
__packaged_file_attrs
--
View it on GitHub:
https://github.com/rpm-software-management/rpm/pull/2734/files/cd4bd81ad1c446c119a27eec77b2828b5ea1624a..5059816af9185cc7c3f19ad729b4dc657f912cde
You are
OK, just to write down how the current system works:
The `fileattrs/*.attr` are read in at the beginning with all other macro files.
The macros in there can be over written by the spec file - or by `%load` ing
the `.attr` file in the Sources.
RPM goes through the list of file attrs (currently
we can use : rpmbuild --without check
--
Reply to this email directly or view it on GitHub:
https://github.com/rpm-software-management/rpm/issues/316#issuecomment-1899391222
You are receiving this because you are subscribed to this thread.
Message ID:
20 matches
Mail list logo