tech-pkg archive

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index][Old Index]

Preliminary status report for the wrapper rewrite

Hi all,
I am currently working on the next generation of the wrapper framework.
I am running a bulk build to find issues with the current status of the
wrappers and the integration. A detailed discussion of the wrappers
themsolve will follow at a later point, bear with me on that one.

This bulk build has currently 246 build failures after 2/3 of the tree
done. There are some high impact packages like cups, but I think the
overview is reasonable.

At the moment 23 packages fail because they depend on the libtool mode
detection. I am still not sure if I want to implement that for the
libtool wrapper or instead fix the breakage. Basically, libtool has some
logic to guess the mode (compile, link, install) based on the command
line. It has also warned about depending on this logic for quite a
while. I have changed (most?) cases where a plain libtool is used in
plain pkgsrc rules, but others remain.

Another 21 packages hit the .buildlink check. I have not checked the
individual cases, especially the shared libraries matching. The idea
behind running the check if the unwrapping sed logic is needed and for
many cases. At the moment it seems like fife packages are affected.

Another 20 packages are broken by stricter checks in the wrapper. At the
moment the wrappers consider it an error if -I, -R and similar options
are missing an argument. This is an explicit error as hunting down the
problems can be rather surprising. This is mostly related to detection
the X11 locations, but there are some other cases.

There are a bunch of yet unclassified errors and a bunch of the normal
bulk build errors that make up the rest.


Home | Main Index | Thread Index | Old Index