Opened 20 years ago

Last modified 20 years ago

#606 closed defect (fixed)

config.* in .cvsignore is not good

Reported by: dougrm@… Owned by: warmerdam
Priority: high Milestone:
Component: default Version: unspecified
Severity: normal Keywords:
Cc:

Description

You can't import a fresh gdal-1.2.1 into a new cvs repository, check it back
out, and configure it.  The config.* in .cvsignore causes the necessary
config.sub and config.guess files to not be put into the repository.  Perhaps
being more explicit with which config.* files to ignore would work, but I'm not
sure what range of files are generated that should be ignored.

Change History (3)

comment:1 by dougrm@…, 20 years ago

One more problem: "man" is inside of .cvsignore.  When you do a "make install"
on a checked-out cvs imported system, the lack of a man directory causes make to
go into an infinite loop.  Very Cool!!  Just creating the directory still makes
"make install" fail because of missing makefile.  Again, the .cvsignore needs to
somehow be sharpened to only ignore newly generated files in man, not the base
files.

Or, maybe the instructions for cvs install could say delete .cvsignore for the
import and then put it back in after first checkout.

comment:2 by dougrm@…, 20 years ago

Nevermind.  I'm learning CVS as I go.  It turns out that the CVS documentation
says (as I suggested, but really thought was silly) its usually a good idea to
remove the .cvsignore files during import and return them after later checkout.
 So this is more a CVS problem/idiom, not a gdal problem.  Though, sharpening
the .cvsignore might possibly to the right thing automatically for CVS-ignorant
people.

comment:3 by warmerdam, 20 years ago

Doug, 

I have modified gdal/.cvsignore to only ignore config.status instead of 
config.*. 

I am not sure that the import issues with .cvsignore files really require
intervention on my part.  The work around would be to remove the .cvsignore
files.  


Note: See TracTickets for help on using tickets.