Show all posts by user
Dear AFNI users-
We are very pleased to announce that the new AFNI Message Board framework is up! Please join us at:
https://discuss.afni.nimh.nih.gov
Existing user accounts have been migrated, so returning users can login by requesting a password reset. New users can create accounts, as well, through a standard account creation process. Please note that these setup emails might initially go to spam folders (esp. for NIH users!), so please check those locations in the beginning.
The current Message Board discussion threads have been migrated to the new framework. The current Message Board will remain visible, but read-only, for a little while.
Sincerely,
AFNI HQ
History of AFNI updates
Page 1 of 2
Pages: 12
Results 1 - 30 of 38
do you remember which version of AFNI you fetched from NeuroDebian which was segfaulting -- may be it is time to let the blood out and update AFNI in NeuroDebian to yet another daily snapshot?
by
yarikoptic
-
AFNI Message Board
I was trying to view some BRIKs which are under git-annex control, thus symlinks to the actual content:
$> ls -l 0back_SD_tbi-hc+tlrc.HEAD
lrwxrwxrwx 1 yoh yoh 197 Jan 5 12:32 0back_SD_tbi-hc+tlrc.HEAD -> ../.git/annex/objects/g8/qf/SHA256E-s2764--57b0b19a22b38f1c78ad6353ed76482e0ab4ab4bfa7f0d9eea8733e6a3daf622.HEAD/SHA256E-s2764--57b0b19a22b38f1c78ad6353ed76482e0ab4ab4bfa7f0d9eea873
by
yarikoptic
-
AFNI Message Board
d'oh -- I should have guessed...
I just want again to express my love to automated tests
by
yarikoptic
-
AFNI Message Board
few days back was introduced to be used in plug_realtime.c but I can't find its definition (seems to be not among afni_pcor.c)... please point to the right direction
by
yarikoptic
-
AFNI Message Board
I wondered if those pubs came out and would appreciate at the pointers at current AFNI way to do preprocessing of the procrastinate-MRI?
by
yarikoptic
-
AFNI Message Board
bloody visual cut paste -- there is a bug somewhere in the system ;)
$> suma -i ld120
Xlib: extension "NV-GLX" missing on display ":2".
vvvvvvvvvvvvvvvvvvvvvvvvvvvvSurface #1/1(Local Domain Parent), loading ...
++ Notice SUMA_LoadSpec_eng (/tmp/buildd/afni-0.20131204~dfsg.1/SUMA/SUMA_Load_Surface_Object.c:4066 @13:32:57):
Have 0 DOs to load
oo Warning
by
yarikoptic
-
AFNI Message Board
I nearly exclusively use SUMA on our shared computing cluster, while running it under TigerVNC (both server and client). Usually currently doing it via the local network (we live at the college housing so the same network as well, before it was through comcast -- also worked nicely even though going somewhere near boston first). A while back I was using custom build of TigerVNC with turbojpeg (
by
yarikoptic
-
AFNI Message Board
Ho ho -- very interesting feature. Reminded me how once while being in Honolulu I have moved the content of /lib away on our cluster in NJ. Few cigarets and beers to cool down put me on the right track then (just in case it would come handy to anyone in similar situation: keep a statically linked/built bash around).
Note that while running as a regular user there should be no detrimental con
by
yarikoptic
-
AFNI Message Board
am I missing any interesting issue? (besides segfaults in suma I am yet to investigate/report on ;) )
by
yarikoptic
-
AFNI Message Board
/home/yoh/deb/gits/pkg-exppsy/afni/SUMA/SUMA_SurfaceMetrics.c: In function ‘main’:
/home/yoh/deb/gits/pkg-exppsy/afni/SUMA/SUMA_SurfaceMetrics.c:831:4: error: format not a string literal and no format arguments [-Werror=format-security]
SUMA_LH (MetricList->s);
^
cc1: some warnings being treated as errors
please patch it up
--- a/SUMA/SUMA_SurfaceMetrics.c
+++ b/SUMA/SUMA
by
yarikoptic
-
AFNI Message Board
heh -- that is on a local computing cluster which runs some aged RHEL... SUMA renders just fine when ran locally so I suspect it is not drivers but GL libraries screwing up what is actually sent to the rendering layer... I am quite ignorant in this topic to deduce which particular layer/library is to blame here
as for remote SUMA pain: well -- have you tried TigerVNC? may be with turbojpeg
by
yarikoptic
-
AFNI Message Board
apparently I was blaming VNC for no good reason. With ssh tunneling it is the same
I guess some of the GL libraries are too aged:
*$> ldd /opt/afni/2011_12_21_1014/suma | awk '/=>.*\//{print $3;}' | xargs ls -l
lrwxrwxrwx 1 root root 11 May 31 17:00 /lib64/libc.so.6 -> libc-2.5.so
lrwxrwxrwx 1 root root 12 May 31 17:00 /lib64/libdl.so.2 -> libdl-2.5.so
lrwx
by
yarikoptic
-
AFNI Message Board
Thanks for addressing this issue! BTW -- you (AFNI hackers) are using an alternative "pattern" for this issue at other places... but suprisingly can't locate it now so I must have dreamed it up... Citing from
This example sprintf() call could be fixed as follows:
-sprintf(buf, "%s plus %d", buf, k);
+sprintf(buf + strlen(buf), " plus %d", k);
Si
by
yarikoptic
-
AFNI Message Board
Julian Taylor (of Debian and Ubuntu fame) suggested to use cppcheck to catch those more reliably than a silly grep... unfortunately it crashed on me at 60% through the code base (choked on count.c) but meanwhile it picked up 1 more:
: (error) Undefined behavior: Variable 'buf1' is used as parameter and destination in sprintf().
It seems to pick up more of problematic cases (some s
by
yarikoptic
-
AFNI Message Board
I had no grep syntax... as for original script the guy used -- it was using google services which are N/A any more:
As for a quick grep line to catch some of the victims (not all since that would require syntax parsing and multiline handling) and not even care about snprintfs which might exhibit similar pattern, this might do it:
$> git grep -e 'sprintf(\s*\(\w\+\)\s*,[^,]\+,\s*\
by
yarikoptic
-
AFNI Message Board
our fellas ran into this issue, which boiled down to finding and reading man 3 printf (search for NOTES and/or undefined). Page even includes grep on afni for points to fix... meanwhile here is the result of such undefined behavior unfortunately with our NeuroDebian build where we use -D_FORTIFY_SOURCE=2 since few months back (I really need to cook up more tests ;) )
$> /usr/lib/afni
by
yarikoptic
-
AFNI Message Board
I have no clue in gentoo but I guess emake is just a 'builder' which could invoke any command to build a package.
To build from our sources smth like this needs to be done
git clone git://git.debian.org/git/pkg-exppsy/afni.git
cd afni
QUILT_PATCHES=debian/patches quilt push -a # will apply all the patches under debian/patches, if no quilt do smth like
# cat debian/patches/serie
by
yarikoptic
-
AFNI Message Board
wonderful! (btw I even tried to place the whole body within code formatting - with the same effect)
I think I had other posts "forbidden" as well without me pointing to any file ... but may be I am mistaken. but next time I will try avoiding pointing to any configuration files under the etc directory
by
yarikoptic
-
AFNI Message Board
Here would be my reply which also got forbidden from being posted
by
yarikoptic
-
AFNI Message Board
Welcome to the "AFNI Builders" club!
My reply could be found on my website because this board has allergy to my persona and claims I have no permission to post (AGAIN and AGAIN) my extended reply (but short notes are ok).
Cheers
by
yarikoptic
-
AFNI Message Board
Tried 2 year old and current builds of AFNI/SUMA under VNC installed on our college cluster -- both exhibit similar behaviour:
by
yarikoptic
-
AFNI Message Board
FWIW -- I have just uploaded a fixed up build of AFNI from 20130830 (-3 Debian revision for that version). The main hurdle ruining builds (and caught by rudimentary tests I added) was to tackle was dealing with --as-needed which is default linkage option on Ubuntu without completely disabling it. It lead to dropping links to gslcblas and then linking against Xt before Xm libraries despite Xm pr
by
yarikoptic
-
AFNI Message Board
Aleks Wrote:
> P.S. Improving suggestion for neuro-debian: Add
> PATH for each installed one automatically!!!
unfortunately this can't happen -- if it could, then we would have simply shipped all AFNI binaries under /usr/bin/.
And the reason is - there is hundreds of cmdline tools productive AFNI developers have brought to public consumption. With such a number and absent u
by
yarikoptic
-
AFNI Message Board
ah -- if it segfaults even with that one -- it works for me with the NeuroDebian build of AFNI sources from 0.20130830
$> 3dSeg -anat TT_N27+tlrc -mask TT_N27+tlrc
++ OpenMP thread count = 2
++ Compile date = Aug 31 2013
3dSeg -anat TT_N27+tlrc -mask TT_N27+tlrc 40,99s user 1,01s system 96% cpu 43,439 total
and even with stock AFNI build from Aug 22
$> /home/soft/afni/lin
by
yarikoptic
-
AFNI Message Board
if you point me to the file in question I could give a shout to a fresh NeuroDebian build to see if issue indeed still persists ;)
by
yarikoptic
-
AFNI Message Board
ha -- but posted a stupid one, I guess I am hitting some kind of 'spam filter' which doesn't like my replies. I wanted to say that you should make sure that your PATH points to afni's binaries
by
yarikoptic
-
AFNI Message Board
atlas colors - tried again with the same "success":
$> ./afni --version
Version AFNI_2011_12_21_1014
[]
I wonder if it could somehow relate to my window manager(awesome running under XFCE4)... I will give it a try on a different box some time later.
linking problem -- it is indeed an odd one since it should have worked (and works for others) if I link to both -lgsl -lgslc
by
yarikoptic
-
AFNI Message Board
Hi Rick,
great to hear that you would get yourself a piece of NeuroDebian running! ;)
I am actually doing final tune ups for our (cmake) build to "robustify" linking and just got stuck on linking problem with Ubuntus:
lesstif -- should be "gone gone" now. Also that issue as I reported happened with stock (your) build as well for me although I always leave a chance
by
yarikoptic
-
AFNI Message Board
Page 1 of 2
Pages: 12