@pmatilai commented on this pull request.
> if (arg == NULL)
return 0;
auto [ ign, val ] = macros().expand_numeric(arg);
-return val;
+if (val > INT_MAX) {
+ rpmlog(RPMLOG_WARNING, _("Macro value too big for int: %" PRIu64 "
Using %i instead.\n"), val, INT_MAX);
> If you are interested, here is my first draft implementation of this:
> https://src.fedoraproject.org/rpms/pyproject-rpm-macros/pull-request/455
This is of course several months late, missed originally due to vacation, but
most certainly have been looking forward to real-world uses, lab testin
Merged #3448 into master.
--
Reply to this email directly or view it on GitHub:
https://github.com/rpm-software-management/rpm/pull/3448#event-15330430320
You are receiving this because you are subscribed to this thread.
Message ID:
___
Rpm-maint mail
@dmnks commented on this pull request.
> snapshot exec "$@" \
- --define "_buildhost testhost" \
- --define "_topdir /build"
+ --define "_tmppath /var/tmp"
This is indeed strange that `runroot` doesn't work with `/tmp` as `_tmppath`,
I've looked at it c
@ffesti commented on this pull request.
> sqlexec(sdb, "PRAGMA optimize");
sqlexec(sdb, "PRAGMA wal_checkpoint = TRUNCATE");
+
+ int max_size = rpmExpandNumeric("%{?_sqlite_vacuum}");
+ if (max_size <= 0)
+ max_size = 20
@dmnks commented on this pull request.
> snapshot exec "$@" \
- --define "_buildhost testhost" \
- --define "_topdir /build"
+ --define "_tmppath /var/tmp"
... but since Ubuntu was now failing, for a change... let's leave it for later,
indeed :smile:
Can't reproduce that here, this is almost certainly some local/distro issue. My
guess is that bzip2 either missing on the system or the macro for it is
pointing to a wrong thing, and in both cases that's something that should be
reported on Mariner, we can't help that (and 4.17-4.18 are out of u
The main commit split into two due to popular demand :smile:
--
Reply to this email directly or view it on GitHub:
https://github.com/rpm-software-management/rpm/pull/3439#issuecomment-2483278884
You are receiving this because you are subscribed to this thread.
Message ID: _
@pmatilai pushed 6 commits.
fd3a7cec0b9278ffd513bb7af966509f5c26ed94 Drop unused header arguments, update
comments in signing internals
05b3bd0beee83e24eeb2854cd4852f8e2941b31b Refactor signature insertion to a
helper function
5cc1cfdaa61fb306bba8832bc0f9471cf2ee28c9 Simplify the v3 signature
@dmnks commented on this pull request.
> snapshot exec "$@" \
- --define "_buildhost testhost" \
- --define "_topdir /build"
+ --define "_tmppath /var/tmp"
Yep, it's tmpfs, but there's only one instance of a podman container for the
whole test-suite. `m
The documentation is at [pyproject-rpm-macros
README](https://src.fedoraproject.org/rpms/pyproject-rpm-macros/blob/rawhide/f/README.md)
-- see the *Provisional: Declarative Buildsystem (RPM 4.20+)* section (sorry,
no anchor).
I generally don't put provisional stuff into the Python guidelines. P
Alas, for your use case, you wanted documentation for Declarative Buildsystems
generally, which is available in
https://rpm-software-management.github.io/rpm/manual/buildsystem.html
--
Reply to this email directly or view it on GitHub:
https://github.com/rpm-software-management/rpm/discussions/
@pmatilai pushed 1 commit.
5834541869bf1c42949113e9d371b0ef46521826 Drop redundant --dbpath from
--exportdb/--importdb test
--
View it on GitHub:
https://github.com/rpm-software-management/rpm/pull/3447/files/9f6c97bc65f8d88cf7ef75c4678a2897ecff90ca..5834541869bf1c42949113e9d371b0ef46521826
Yo
I just stumbled over this as
[five](https://discussion.fedoraproject.org/t/should-we-check-for-the-presence-of-install-in-spec-files/137316)
package already use this new "BuildSystem: pyproject" approach but I can't
find any official documentation. If this is already in use and accepted for
Fed
@pmatilai commented on this pull request.
> snapshot exec "$@" \
- --define "_buildhost testhost" \
- --define "_topdir /build"
+ --define "_tmppath /var/tmp"
It's tmpfs though, I'd think it gets cleared between test-groups. At least it
does between "ma
Thanks, I was searching for the meaning of "BuildSystem: pyproject" and didn't
see anything from the Fedora domains so this discussion here looked the most
interesting on the search results. Your first link is what I was looking for at
the moment, saying that it is new, provisional and able to r
@pmatilai commented on this pull request.
> snapshot exec "$@" \
- --define "_buildhost testhost" \
- --define "_topdir /build"
+ --define "_tmppath /var/tmp"
Eh, or that's what I thought, but apparently not.
--
Reply to this email directly or view it
@pmatilai commented on this pull request.
> @@ -683,3 +683,20 @@ runroot rpm -q 'versiontest-1.0~2-1'
[])
RPMTEST_CLEANUP
+
+# --
+AT_SETUP([rpmdb vacuum])
+AT_KEYWORDS([install rpmdb sqlite])
+cp -r /usr/lib/sysimage/rpm "${RPMTEST}"/tmp
+RPMTEST_CHECK([
+runroot
@pmatilai commented on this pull request.
> snapshot exec "$@" \
- --define "_buildhost testhost" \
- --define "_topdir /build"
+ --define "_tmppath /var/tmp"
The problem is that /var is read-only by default, and rpm needs temporary files
for all sorts
Note, there is a pattern here that seems valid to me, which triggers this
warning. See [Fedora's evolution-data-server
package](https://src.fedoraproject.org/rpms/evolution-data-server/blob/rawhide/f/evolution-data-server.spec).
An edited extract:
```
# 0-99: General patches
# 100-199: Flatpak
The question is what to do with rpmExpandNumeric() that uses this interface. It
needs to have defined behavior in case of overflow, but there's no way to
return an error from rpmExpandNumeric().
--
Reply to this email directly or view it on GitHub:
https://github.com/rpm-software-management/rpm
There are about a thousand further cases were runroot() can be eliminated but
lets leave something for another rainy day :sweat_smile:
--
Reply to this email directly or view it on GitHub:
https://github.com/rpm-software-management/rpm/pull/3447#issuecomment-2482254875
You are receiving this be
I suppose the least dangerous option is to emit a warning and cap the return
value to int min/max.
--
Reply to this email directly or view it on GitHub:
https://github.com/rpm-software-management/rpm/pull/3454#issuecomment-2482891022
You are receiving this because you are subscribed to this thre
Closed #3435 as completed via 35cfb7de9842dbec35adcb9ad9a632df49f35102.
--
Reply to this email directly or view it on GitHub:
https://github.com/rpm-software-management/rpm/issues/3435#event-15330430599
You are receiving this because you are subscribed to this thread.
Message ID:
__
@ffesti pushed 1 commit.
8fca2a2eea43e2667b01c541ac3eeae624f19e1c Make macros::expand_numeric return
int64_t
--
View it on GitHub:
https://github.com/rpm-software-management/rpm/pull/3454/files/c70c9406eef823eda258e6327fe72142a8fb46cf..8fca2a2eea43e2667b01c541ac3eeae624f19e1c
You are receiving
@pmatilai pushed 2 commits.
883b132422ace3e5b04619a149091e9bf3c7772c Eliminate another bunch of now
unnecessary runroot() uses
3989f576e232e968ab457e7f9d19fb0a6b057fae Eliminate bunch of now unnecessary
runroot_other() uses
--
View it on GitHub:
https://github.com/rpm-software-management/rpm
Check if there is 20MB in free pages and then execute a VACUUM command. The
threshold can be controlled by the _sqlite_vacuum macro. Don't add this to
the macros file on purpose as we don't want people to get involved with
such details. Here it is mainly used for testing.
Using a 20 MB threshol
As for the "Ubuntu mystery", see
https://github.com/rpm-software-management/rpm/pull/3447/commits/06e4da15fc61e6366f21f992f4a2b6af428df3a8
- I still don't know why exactly its failing for that one path but not for
something else that should be on the same image, but for the purpose of that
test
@dmnks commented on this pull request.
> snapshot exec "$@" \
- --define "_buildhost testhost" \
- --define "_topdir /build"
+ --define "_tmppath /var/tmp"
Yep, I realized that `/var/tmp` shouldn't be writable after writing the above
comment, too :sweat
@dmnks commented on this pull request.
> snapshot exec "$@" \
- --define "_buildhost testhost" \
- --define "_topdir /build"
+ --define "_tmppath /var/tmp"
Oh, right. So `/var/tmp` is r/w even in `podman --read-only` mode, see the man
page for `--read-o
@dmnks commented on this pull request.
> snapshot exec "$@" \
- --define "_buildhost testhost" \
- --define "_topdir /build"
+ --define "_tmppath /var/tmp"
So using `/var/tmp` would work for the "native" (originally `run`) case, too,
however it would vi
@ffesti pushed 1 commit.
5e674ede9f0bf42f0a6b8976969f40bf4d991405 Vacuum the sqlite rpmdb if necessary
--
View it on GitHub:
https://github.com/rpm-software-management/rpm/pull/3452/files/190ef17e7427fae1ec08f4757299a2a832eb7cfc..5e674ede9f0bf42f0a6b8976969f40bf4d991405
You are receiving this b
@pmatilai commented on this pull request.
> snapshot exec "$@" \
- --define "_buildhost testhost" \
- --define "_topdir /build"
+ --define "_tmppath /var/tmp"
That's handy, if it actually works for us. It seems to, locally, whereas /tmp
didn't. I'll pus
@pmatilai pushed 1 commit.
fe49d958024126dcde01e442e9745a44b4cc44a4 Always use /var/tmp for rpm's
%_tmppath in test-suite
--
View it on GitHub:
https://github.com/rpm-software-management/rpm/pull/3447/files/3989f576e232e968ab457e7f9d19fb0a6b057fae..fe49d958024126dcde01e442e9745a44b4cc44a4
You
@pmatilai pushed 0 commits.
--
View it on GitHub:
https://github.com/rpm-software-management/rpm/pull/3447/files/fe49d958024126dcde01e442e9745a44b4cc44a4..3989f576e232e968ab457e7f9d19fb0a6b057fae
You are receiving this because you are subscribed to this thread.
Message ID:
___
@pmatilai commented on this pull request.
> snapshot exec "$@" \
- --define "_buildhost testhost" \
- --define "_topdir /build"
+ --define "_tmppath /var/tmp"
And, it may be writable on podman but apparently not docker. I'll just unpush
the last commit,
@dmnks commented on this pull request.
> snapshot exec "$@" \
- --define "_buildhost testhost" \
- --define "_topdir /build"
+ --define "_tmppath /var/tmp"
That said, I've just realized we've been doing it previously the same way,
`run` used to override
Merged #3447 into master.
--
Reply to this email directly or view it on GitHub:
https://github.com/rpm-software-management/rpm/pull/3447#event-15333249890
You are receiving this because you are subscribed to this thread.
Message ID:
___
Rpm-maint mail
Looks good, well done :smile: The courage for diving into this bottomless pit
is commendable (and very much appreciated) :smile:
--
Reply to this email directly or view it on GitHub:
https://github.com/rpm-software-management/rpm/pull/3447#issuecomment-2482831336
You are receiving this because
@pmatilai converted this issue into discussion #3453.
--
Reply to this email directly or view it on GitHub:
https://github.com/rpm-software-management/rpm/issues/3449#event-15331099004
You are receiving this because you are subscribed to this thread.
Message ID:
@ffesti pushed 1 commit.
190ef17e7427fae1ec08f4757299a2a832eb7cfc Vacuum the sqlite rpmdb if necessary
--
View it on GitHub:
https://github.com/rpm-software-management/rpm/pull/3452/files/a41e2b262470a33bdbb7a514e840959edba65f5b..190ef17e7427fae1ec08f4757299a2a832eb7cfc
You are receiving this b
While we can't change the C API there is really no reason to add a new 32
bit API. Changing now while it is still internal.
You can view, comment on, or merge this pull request online at:
https://github.com/rpm-software-management/rpm/pull/3454
-- Commit Summary --
* Make macros::expand_num
@ffesti commented on this pull request.
> @@ -683,3 +683,20 @@ runroot rpm -q 'versiontest-1.0~2-1'
[])
RPMTEST_CLEANUP
+
+# --
+AT_SETUP([rpmdb vacuum])
+AT_KEYWORDS([install rpmdb sqlite])
+cp -r /usr/lib/sysimage/rpm "${RPMTEST}"/tmp
+RPMTEST_CHECK([
+runroot r
Ehm.. only saw the mention of squashing half a second too late.
--
Reply to this email directly or view it on GitHub:
https://github.com/rpm-software-management/rpm/pull/3448#issuecomment-2482296309
You are receiving this because you are subscribed to this thread.
Message ID: ___
@pmatilai commented on this pull request.
> snapshot exec "$@" \
- --define "_buildhost testhost" \
- --define "_topdir /build"
+ --define "_tmppath /var/tmp"
Right. Anyway, lets leave that to a future exercise :laughing:
--
Reply to this email direct
@pmatilai commented on this pull request.
> snapshot exec "$@" \
- --define "_buildhost testhost" \
- --define "_topdir /build"
+ --define "_tmppath /var/tmp"
Right, in that case. We could of course just rm -rf the contents at
RPMTEST_CLEANUP or such, b
@pmatilai commented on this pull request.
> sqlexec(sdb, "PRAGMA optimize");
sqlexec(sdb, "PRAGMA wal_checkpoint = TRUNCATE");
+
+ int max_size = rpmExpandNumeric("%{?_sqlite_vacuum}");
+ if (max_size <= 0)
+ max_size =
47 matches
Mail list logo
