trunk/Open-ILS/src/c-apps/Makefile.am has an error

[Steven Chan]

Changeset 11550 did this:

Changed:
oils_cstore_la_LDFLAGS = $(AM_LDFLAGS) -loils_idl -ldbi -loils_utils -ldbdpgsql -module
oils_cstore_la_DEPENDENCIES = liboils_utils.la liboils_idl.la

To this:
oils_cstore_la_LDFLAGS = $(AM_LDFLAGS) -loils_idl -ldbi -ldbdpgsql -module
oils_cstore_la_DEPENDENCIES = liboils_idl.la

Changeset 11551 tried to reversed parts of 11550, but made an error:

Changed:
oils_cstore_la_LDFLAGS = $(AM_LDFLAGS) -loils_idl -ldbi -ldbdpgsql -module
oils_cstore_la_DEPENDENCIES = liboils_idl.la

To this:
oils_cstore_la_LDFLAGS = $(AM_LDFLAGS) -loils_idl -ldbi -ldbdpgsql -loils_utils -module
oils_cstore_la_DEPENDENCIES = liboils_idl.la liboils_idl.la

[Dan Scott (denials)]

To be clear, the reported problem is the double-entry of liboils_idl.la instead of the addition of liboils_util.la to the oils_*store_la_DEPENDENCIES lines.

And yet... the thing builds and runs, and has for over a year. So a little more investigation is required.

[Jason Stephenson (jstephenson)]

Is this still valid?

[Steven Chan (schan2)]

The oddity of the double-entry of liboils_idl.la in the DEPENDENCIES line was reported about a year ago, and around that time, I did a shallow investigation of the Evergreen makefile system and found it to be written in a 'nonstandard' way. I got detoured into other work and never returned to fleshing out the problems.

I agree that the makefiles work as is, but the double entry hints at problems that will occur if a developer tries to make targets that are expected to work, but do not. I can review my year-old notes and make further comments.

[Michael Peters (mrpeters)]

I've verified and this doesn't appear to be incorrect in master anymore, therefore I'm closing this bug.

The following commit fixed this:

http://git.evergreen-ils.org/?p=Evergreen.git;a=commitdiff;h=3f914e81827a5bc9682dfbfdd16eaec571281505