https://bugzilla.redhat.com/show_bug.cgi?id=658754
--- Comment #12 from Esen Sagynov <esen.sagy...@nhn.com> ---
Hello Cristian,
First of all, thank you for your continuos support on IRC. You feedback was
very valuable for us to improve our CUBRID spec.
Below please find our last changes we've made according to your comments.
1) Regarding macros, we've replaced %{_prefix}/share/ with %{_datarootdir} and
%{_prefix}/include -> %{_includedir}
2) "Some files under /usr/include and share are owned by cubrid even if they
shouldn't." /usr/include directory is not owned by cubrid user. Only share dir
was. But when we tried to chown root:root, we found out that cubrid master
service checks if the files are owned by the user who is running the process.
When root:root, the cubrid master service fails to load saying “Current user
does not match CUBRID user”. Therefore, we had to roll back and chown
cubrid:cubrid the shared files. We have reported this issues in our Issue
Tracker, so we will look into it in the future version.
4) Regarding "devel-file-in-non-devel-package", like I explained before some
libs must be in the main package because CUBRID uses them, otherwise, CUBRID
will not work; we moved all libs to devel package which were possible to move.
5) Regarding, "put all commands under one giant command command like git does",
the "cubrid" command actually does exactly that. This is that giant command.
6) “strip --strip-unneded”: we have completely removed strip commands, which
now gives us only warnings saying "W: unstripped-binary-or-object".
7) This allows us at least have the debuginfo package created and it has no
errors related to "empty-debuginfo-package".
8) We now keep changelogs up to date. Also spec release number is always UP'ed
and file version is updated if necessary.
9) According to your suggestion, we added conditionals to handle different Java
versions for different versions of Fedora.
10) For new Fedora we use systemd now, and for old including RH distributions
use "SysV scripts".
11) "Replace %{libdir} with the standard %{_libdir} macro". As I explained
before, we cannot do this in this version because we have a hardcoded /usr/lib
path which is used even on 64-bit system architecture. We will fix this in the
CUBRID core in the future version.
12) We no longer start the program/service after installation.
13) "If possible, use a private /tmp in your systemd service unit". We didn't
add "PrivateTmp=true" because "cubrid master" process failed to start if this
line was set in cubrid.service.
I am glad that so far CUBRID package is being successfully built at Fedora.
Please let me know if there is anything we can do more to improve the CUBRID
spec.
Thank you Cristian!
Regards,
Esen Sagynov.
--
You are receiving this mail because:
You are on the CC list for the bug.
_______________________________________________
package-review mailing list
package-review@lists.fedoraproject.org
https://admin.fedoraproject.org/mailman/listinfo/package-review
