BIND 10 #1501: [kean] Make install fails with relinking of sqlite3 datasource backend
BIND 10 Development
do-not-reply at isc.org
Tue Nov 26 05:23:22 UTC 2013
#1501: [kean] Make install fails with relinking of sqlite3 datasource backend
-------------------------------------+-------------------------------------
Reporter: vorner | Owner:
Type: defect | UnAssigned
Priority: high | Status:
Component: build system | reviewing
Keywords: | Milestone:
Sensitive: 0 | Sprint-20131015
Sub-Project: DNS | Resolution:
Estimated Difficulty: 5 | CVSS Scoring:
Total Hours: 0 | Defect Severity: N/A
| Feature Depending on Ticket: none
| Add Hours to Ticket: 0
| Internal?: 0
-------------------------------------+-------------------------------------
Changes (by kean):
* status: new => reviewing
Comment:
This is very easily reproduced and very difficult to fix. It is a
fundamental problem with automake. The problem is that "make install" only
has dependencies on the target it is installing, not on other components.
I am not aware of any mechanism for introducing dependencies for the
install rule for generated makefiles. This is further complicated by
libtool which when installing does a relink to get the correct "final"
SONAME entries into the shared abject. Doing parallel installs is bad for
other reasons too. You can end up with corrupted archives if you have
multiple sub-makes all adding members to a ranlib archive.
However, it is worth noting that using the jobserver for "make install" is
highly unconventional for the above reasons, and this problem and be very
easily avoided by simply not doing that. At best this warrants some
documentation in a README or the building instructions.
--
Ticket URL: <http://bind10.isc.org/ticket/1501#comment:14>
BIND 10 Development <http://bind10.isc.org>
BIND 10 Development
More information about the bind10-tickets
mailing list