[gpm]Re: new release: gpm-1.19.5
Nico Schottelius
nicos@pcsystems.de
Thu, 27 Sep 2001 07:29:10 +0200
Hello guys!
Yesterday I finished most of the bugs.
I try to fix the others this weekend. Time is running away
currently and right now it is 1.43am again.
I will reply to this mail again, with comments to each bug.
Nico
Alessandro Rubini wrote:
> > Once again there is a new release of GPM!
>
> It has a few bugs added. Some may be older (like 1.19.4 that I didn' look at).
>
> * After running ./configure in the "make" it re-runs autoconf and configure.
> autoconf may be not installed in the user's computer.
>
> * Tries to link gpm with libgpm. That's not needed. The resulting binary
> looks linked to lgpm (this means that if you have no libgpm in
> your library path the newly compiled daemon won't run
>
> * Tries to make t-mouse.el in src/ , but it's in contrib/emacs
> (not that it's not a contrib source, it's a core gpm feature
>
> * The various "ex_tex" etc stuff is bugged, comparison always fails
> (no $$ in Makefile)
>
> * Liblow.c defines non-static vcname and consolename. It pollutes application
> name space. Same for check_devfs and set_devfs, should be static into
> liblow.c
>
> * src/Makefile: LSRC should list the lib sources, new stuff is added to
> LOBJ instead (no bug, just unclean)
>
> * liblow.c::Gpm_Open: the comment asks what's isdigit() needed: it's in
> the next line, where you set conn->vc=atoi() .
>
> * mouse_test.c has the buggy check_devfs: opened fd is not closed, not opened
> is closed.
>
> * This in gpm.c:
> chmod(GPM_NODE_CTL,0700);
> allows only root to connect to the server
>
> * The new headers don't protect against double-inclusion
>
> * defines.h replicates a lot of stuff from gpm.h
>
> * ChangeLog doesn't use the right format. It should list what has been
> changed in each file (otherwise people can't look for a specific
> file that they are interested in)
>
> > o new daemon name
>
> Sorry? Didn't find it.
>
> /alessandro "finding bugs is easier than maintaining a program"