@jon13 would you please consider committing the attached patch?
If I recall correctly having rpcbind running is a requirement.
ksh93: import 1.0.4
Reject are being caused by https://sourceforge.net/p/cdesktopenv/code/ci/ef2c655f552b3ca3210a5580993057937fa8fa7c/ https://sourceforge.net/p/cdesktopenv/code/ci/76812a9115db584cb4754761f46e9c76081d2435/ Which are now fixed upstream
After some more experimentation: Patching to 1.0.4 seems possible with no rejects. Patching up to 1.0.8 seems possible with only 2 rejects (in xec.c)... Patching from 1.0.9 introduces 16 rejects (in aso)...
As an aside, I did make a diff of ksh-1.0.3 and ksh-1.0.10 and tried applying that to CDE, which ended up with rejects in src/cmd/ksh/sh/xec.c and src/lib/libast/features/aso.
with regard to dtcodegen any calls to CATGETS seem to trigger a Segmentation Fault...
The attached v4.2 patch also fixes 'dtlogin' so that works now too...
The attached v4.2 also fixes 'dtlogin' so that works now too...
The plot thickens, when executing dtcodegen without NLSPATH= it seems to work just fine... Unrelated Sidenote: programs/dtimsstart/main.c contains a reference to ENV_NLSPATH, wonder if that actually intentional?!?!?
without dtappbuilder and ttsnoop, startx /usr/dt/bin/dtsession starts just fine. @jon13 would you please consider applying my v3 patch, given that those changes seem to be fine, even if they don't solve all issues yet.
So after removing dtappbuilder and ttsnoop from programs/Makefile and compiling with my v3 patch applied the rest of CDE builds just fine.
And this is still reproducible with gcc 13.4...
And this is still reproducible with gcc 13.4...
@jon13 I'm a bit at a loss with regard to the dtcodegen issue, any insights on your end?
@jon13 I'm a bit at a loss with regard to the dtcoregen issue, any insights on your end?
I've tested ksh93 outside of cde, and both 1.0.3 AND 1.0.10 still have the readonly.exe signal 11... So this isn't a CDE issue, but likely ksh93 upstream, and not fixed yet in the latest release...
ksh readonly.exe exited on signal 11
ksh readonly.exe exited on signal 11
However dtcodegen is segfaulting now... NLSPATH=../../../../programs/dtappbuilder/src/abmf/dtcoregen.cat ../abmc/dtcodegen -changed -merge -p dtbuilder.bip about_box.bil See attached backtrace... I'm unsure what's happening here...
./configure --with-libiconv-prefix=/usr/local fixed that issue. I solved a few more minor issues, see v3 patch attached.
So the UNIXbindings issue was just me not using gmake ... I've managed to make some further progress, see attached patch... Now I'm stuck at : RFCBodyPart.C: error: undefined reference to 'libiconv_open' RFCBodyPart.C: error: undefined reference to 'libiconv' RFCBodyPart.C: error: undefined reference to 'libiconv_close'
With the attached patch applied, the build gets stuck at: rm -rf UNIXbindings ../../util/tradcpp/tradcpp -P -DXCOMM=# -DCDE_INSTALLATION_TOP=/usr/dt -I../../include/Dt > UNIXbindings Any hints?
CDE fails to build on DragonFly
I just noticed uncompress is checked for as well, but I don't think it's actually used?
_FORTIFY_SOURCE=3 dtksh-init and tirpc
configure doesn't check for (n)compress
Ideally cdesktopenv shouldn't depend on compress any more... both lz4 and zstd would be excellent replacements, as they are both faster, and compress better... but I realize the decompression code in CDE would probably be nontrivial to replace.
Apparently ./configure doesn't check for it...
Aaaah, I was missing ncompress....
Briefly I thought it might be due to mksh, which I replaced with proper ksh93, but no change... Another thought that occurred is that Alpine uses Busybox for it's main userland, any clue whether there's any dependencies on GNU behaviors that the CDE build system might rely on?
gencat musl libc
opensp-1.5.2 is installed
onsgmls: duplicate declaration of parameter entity "local.notations"
CFLAGS=-std=gnu17 may provide short term workaround...
CFLAGS=-std=gnu17 may provide short term workaround...
The patch seems to work, the build continues...
cm_i18n.c:394:21: error: implicit declaration of function 'fcvt'
CAM: fix PocketShape OffsetPattern broken by #23649
CAM: Pocket sensible default StepOver
CAM: PocketShape defaults to Offset and 50% StepOver
PartDesignTests: explicitly thread pitch for regular profile ISO threads, fix enum
And as proper git patches see new attachments.
FLTK_CONFIG variable for cross-compile
0002 is actually a bugfix though...
Superseded by #33, please close.
Superceded by #33, please close.
ui/prefs additions
Sensible default custom gui_custom_bg color
Eureka Debian Patches Upstream
Expand FreeDM/FreeDOOM2 support
OpenBSDBuild
OpenBSDBuild
OpenBSDBuild
OpenBSDBuild
OpenBSDBuild
OpenBSDBuild
OpenBSDBuild
Possibly something along these lines would need to be added: https://cvsweb.openbsd.org/cgi-bin/cvsweb/ports/x11/slim/patches/patch-app_cpp?rev=1.6&content-type=text/x-cvsweb-markup
OpenBSDBuild
OpenBSDBuild
issue #64 is not resolved by the above patches
OpenBSD 6.2 build fixes
Possibly something along these lines would need to be added to dtlogin: https://cvsweb.openbsd.org/cgi-bin/cvsweb/ports/x11/slim/patches/patch-app_cpp?rev=1.6&content-type=text/x-cvsweb-markup
OpenBSDBuild
OpenBSDBuild
OpenBSDBuild
OpenBSDBuild
OpenBSDBuild
OpenBSDBuild
OpenBSDBuild
OpenBSDBuild
OpenBSDBuild
OpenBSDBuild
OpenBSDBuild
OpenBSDBuild
This caused #56 and #57, so it didn't completely fix the issue, and apparently went...
This caused #56 and #57, so it didn't completely fix the issue, and apparently went...
This should fix it.
This happened because of #52, particularly https://sourceforge.net/p/cdesktopenv...
With some extra guidance you provided via IRC, I've got dtlogin working again, with...
dtlogin segfaults on openbsd 5.8
[PATCH] Various new camera models