Hi,
I decided to put my debugging scripts into git, scripts/
directory.
This will scale better than exchanging them with all of you
one at a time.
They all meant to be run from a top of a scratch copy of the tree.
Here they are:
scripts/dbg_mkrpm -
Builds new rpm's from source and leaves them in the top of the tree.
scripts/dbg_rpmbuildlocal -
Helper for the above. I use it for many projects
and on my box, it sits in ~/bin
scripts/dbg_rpminst -
Deinstalls all installed packages with word "abrt" in them,
then installs all *.rpm it sees in current directory.
Used to install new rpm's produced by dbg_mkrpm run.
Very dumb: deinstalls/reinstalls in the loop until
succeeds - "poor man's dependency tracking".
If you know how to fix that, please let me know.
scripts/dbg_unpkrpm -
Unpacks all *.rpm it sees in current directory unto UNPACKED/*.
Useful if you want to peek into rpm's contents.
These scripts are crude in places, feel free to adapt to your needs.
I know that Jiri installs his test builds into his home directory.
IOW, he builds the source with configure --prefix=SOMETHING.
I was also doing something like this, but stopped doing it.
The rationale is this way my test builds were different from
what package users will have. Therefore I'll see some bugs
they will never see, and vice versa - I will fail to catch
some bugs which happen only with standard build, so these
bugs will go full round - package is released, users hit bugs,
users report them. I spend lots of time reproducing their bugs
(since they dont happen to me in non-std build).
In short:
Currently, we need to reduce the number of problems
with standard build.
We can debug problems with custom builds later.
--
vda
Show replies by date