<megatorus>
guess it was reintroduced? maybe only under certain conditions?
<megatorus>
just wondering if i should go to the trouble of signing up for mantis to file a new bug on this, or maybe someone can double-check on this for me before i do?
<thelema>
megatorus: where are your x11 libs?
<thelema>
rixed: git HEAD?
<thelema>
rixed: I need to tweak the makefile some more - my fault.
ftrvxmtrx has joined #ocaml
larhat has quit [Quit: Leaving.]
<megatorus>
thelema: /usr/X11R6/lib64 is where i ended up pointing ./configure with --x11lib for it to work.
<megatorus>
without it, it was looking in /usr/X11R6/lib
<rixed>
thelema: now at 092f2b0e286edce8dc5e773505e3bbb723e775d9
<thelema>
rixed: yup, that's the current head. I made a tweak to doc building and apparently didn't fix the doc installation to match
<thelema>
rixed: try c34f7ed, just pushed to github
<thelema>
err, one more fix needed
<thelema>
rixed: ok, ba5df13
<megatorus>
by the way, i note that the caml site doesn't list a git repo, only svn. is development now being done on git, as seems to be the case based on your present conversation?
<thelema>
megatorus: our conversation is regarding batteries. ocaml development is in svn
<megatorus>
ah ok. got my hopes up for nothing, then. :)
<megatorus>
so any thoughts on the x11 libs issue?
<thelema>
I remember fnding a list of lib search dirs in the configure script; did you use configure?
Kakadu has joined #ocaml
<megatorus>
yes i did, as noted above. my first attempt was ./configure --prefix <somepathinmyhomedir>
<megatorus>
when that failed due to wrong x11 libs, i added --x11lib /usr/X11R6/lib64
<megatorus>
and that worked
<thelema>
hmm, I don't think that configure is meant for win32... checking the docs...
<megatorus>
this is on linux
<thelema>
oh, hmm. I must have misread again, somehow I thought you were on win32
<thelema>
ok, time to check configure
ocp has joined #ocaml
<megatorus>
this is on a university compute box that i have no root on, hence my --prefix to install in home dir. :)
<adrien>
that's one my system btw: lrwxrwxrwx 1 root root 5 Apr 29 2011 X11 -> X11R6/
<megatorus>
it's a suse 9
<adrien>
megatorus: old box?
<megatorus>
well, not that old.
<adrien>
a few years
<adrien>
?
<megatorus>
yea, a couple years i guess
<thelema>
very odd, /usr/X11R6/lib64 looks to be the first directory to be scanned (after $x11_lib_dir and $x11_try_lib_dir). examining further
<thelema>
$x11_lib_dir is set by --x11lib
<megatorus>
whenever SUSE enterprise 9 was released, adrien :)
<thelema>
hmm, where's your x11 include dir?
Drakken has joined #ocaml
<thelema>
ah, that's the problem. $x11_try_lib_dir is $x11_include =~ /include/lib/
<thelema>
meaning it detects your include dir, and then tries something based on that.
<megatorus>
include is in /usr/X11R6/include
<megatorus>
yea
<thelema>
which means it tries /usr/X11R6/lib first...
<thelema>
ok, it just needs to try s/include/lib64/ as well.
<thelema>
do you understand well enough to submit the bug to mantis?
Cyanure has joined #ocaml
<megatorus>
yea i can do it, though signing up for "yet another website (tm)" is a pain ;)
<megatorus>
rixed: no, that was the break between classes
<megatorus>
i'm in another class now
protz has joined #ocaml
protz has left #ocaml []
ankit9 has quit [Ping timeout: 245 seconds]
raichoo has quit [Quit: Lost terminal]
oriba has joined #ocaml
ppm has joined #ocaml
ppm has quit [Client Quit]
jonludlam has quit [Remote host closed the connection]
megatorus has quit [Quit: Page closed]
eikke has quit [Read error: Operation timed out]
Kakadu has quit [Quit: Konversation terminated!]
srcerer_ is now known as srcerer
ikaros_ has quit [Quit: Ex-Chat]
ftrvxmtrx_ has joined #ocaml
<thelema>
can someone test `ocaml odb.ml --unstable lablgtk2` ?
<thelema>
biab
ocp has quit [Ping timeout: 240 seconds]
<edwin>
thelema: works, but complains that /home/edwin/.odb/lib is not in my OCAMLPATH
<edwin>
if I run as root, can it install to /usr/local though?
<edwin>
$HOME/.odb looks like a weird place to install packages to
<_habnabit>
why's it weird? just set OCAMLPATH=$HOME/.odb/lib and run it again
<_habnabit>
the default reaction to "a thing didn't work" shouldn't be "try to run it again as root"
<edwin>
thats why I asked before running as root ;)
<edwin>
Successfully installed lablgtk2 with OCAMLPATH set
<thelema>
edwin: --sudo
Cyanure has quit [Read error: Operation timed out]
<edwin>
I think I'm just used to C libraries installing to /usr/local by default (and complaining I'm not root). And I have to set a --prefix $HOME/something to get it installed where I wanted
<thelema>
edwin: or as root with --have-perms
<edwin>
can I tell it to install to $HOME/ocaml for example?
<edwin>
odb.ml --help doesn't list --prefix
roha has joined #ocaml
<thelema>
for that, just edit the odb.ml file and replace "odb_home" with wherever you want.
<edwin>
:)
<thelema>
I didn't make it configurable, as I don't want too many config options
<thelema>
I already have too many
<edwin>
also is there support for checking digital signatures on the downloaded source packages?
<thelema>
no, but there's no signing infrastructure, so...
<edwin>
ah so oasis-db has to be taught to do that first
<thelema>
edwin: is that really a big deal? Anyone can upload a package, so having security between oasis-db and odb wouldn't help
<edwin>
well there could be a special keyring package with trusted maintainers
yezariaely has joined #ocaml
<edwin>
or the user could manage that on his own
yezariaely has left #ocaml []
<edwin>
of course this will complicate the user experience, hmm
<thelema>
edwin: my opinion: it's much more work than it's worth
<edwin>
yeah, but I don't like grabbing random things from the internet and executing them, especially if its all coming from the same server
<thelema>
edwin: granted. But when was the last time you audited a makefile?
<edwin>
well I'd rather trust the authors of a package to do that. So lets say I trust the maintainer of lablgtk2 to not do bad things. then even if oasis-db is compromised and a bad lablgtk2 is uploaded the digital signature check would fail
<edwin>
oasis-db would only need to support uploading an additional file: .sig with each package
<edwin>
which is a GPG signature made by the author, with his/her own key
<thelema>
and oasis-db would also have to have a database of these keys
<thelema>
and an asociation between keys and projects
<thelema>
so that the right projects would be signed by the right authors
<edwin>
a simplified key verification scheme would be: first time you see new key (in odb) you ask the user if they want to trust it, if so you import it into an odb-trusted.gpg keyring. The next time that author uploads a package, it will be trusted
<thelema>
and users will always trust keys, especially if I add a key that says I'm "Xavier Leroy"
<thelema>
yes, this moves the blame to the user for agreeing to trust that key, but it doesn't really solve any problem.
<edwin>
hmm, well the other way is to do it the Debian way
<thelema>
yes, key signing events at the ocaml meetings
<thelema>
I wouldn't ever ask the user, but would move that responsibility to the repo maintainers to verify packages
<edwin>
there would be a keyring package, with project - pubkey of uploaders
<thelema>
clearly the user trusts the repo maintainer.
<edwin>
and then there would be a requirement like: all packages in distribution N have to verifiable using the keyring package from version N. and the keyring package from version N+1 has to be verifiable by the keyring in version N
<thelema>
and at the repo level, since oasis-db already tracks the uploader, it's just a matter of noting who uploaded each package.
<thelema>
no public key anything.
<edwin>
still, the responsibility of the repo admin would be huge
<thelema>
to verify each package?
<thelema>
no, just to verify each uploader
<edwin>
the uploaders
<edwin>
I think pubkey is still needed so that the client (odb.ml) can still verify the packages, in the event that the oasis-db server is compromised
<edwin>
and after the kernel.org compromise I don't consider compromise of project server X unlikely :(
<thelema>
Extra C object files: -L/usr/lib/x86_64-linux-gnu -lpcre -lpcre_stubs
<thelema>
grr, wrong order...
sepp2k has joined #ocaml
<thelema>
grr, OCamlMakefile is bugged
<sgnb>
gerd's idea actually crossed my mind... but I find 2 weeks quite optimistic
<thelema>
sgnb: 2 weeks to build on the shoulders of giants, maybe.
<thelema>
edwin: would you be happier if odb downloaded directly from VCSes?
<edwin>
wouldn't make much difference, then we'd need to check the signature on git tags
ocp has joined #ocaml
pilki has joined #ocaml
<edwin>
thelema: I'll open a feature request when I made up my mind how the signing scheme should look like, give me a few days to think about it
<edwin>
should be something thats simple, and doesn't take up too much time to implement, and too much time to use on a daily basis
<edwin>
will probably be somehting based on gpg, and signing a list of public keys that are valid for each project, and have the project admin's key signed by the oasis-db admin's
<thelema>
ok, think hard
sebz has joined #ocaml
<thelema>
awesome - right as I go to submit the bug, I check the repo and OCamlMakefile has already fixed it. Go Markus
ocp1 has joined #ocaml
ocp has quit [Quit: Leaving.]
<thelema>
after a 5 hour detour, I'm finally able to install benchmark to test root1d
<bitbckt>
the yaks are restless, today.
<thelema>
bitbckt: well, I'm sorta responsible for odb, and *many* source packages are broken by linux's linking order change.
<bitbckt>
"sorta" here meaning "are" :P
<thelema>
well, I'm responsible for odb, I don't *have* to take responsibility for every package uploaded to oasis-db
<bitbckt>
yup
ocp1 has quit [Read error: No route to host]
ocp has joined #ocaml
Snark has quit [Quit: Quitte]
<roha>
hey, which book would you recommend: introduction to ocaml by jason hickey or ocaml for scientists by john harrop, or maybe something else?
<thelema>
roha: browsing hickey's book, it seems good. I've no copy of harrop's book, so can't compare.
<thelema>
although to be honest, I learned the most from re-re-re-reading the ocaml manual.
thomasga has quit [Quit: Leaving.]
gdu has joined #ocaml
<roha>
ok, thanks. ill think ill go with hickey's book and the ocaml manual then. probably it doesn't matter that much anyway, since i first have to get my head wrapped around the whole functional programming way of thinking :)
<thelema>
yup, that'll take a little while - it's hard to train yourself away from mutating values all the time
ocp has quit [Quit: Leaving.]
ocp1 has joined #ocaml
ocp1 has quit [Ping timeout: 240 seconds]
ocp has joined #ocaml
ocp has quit [Write error: Broken pipe]
ocp1 has joined #ocaml
sebz has quit [Quit: Computer has gone to sleep.]
ocp1 has quit [Client Quit]
ocp has joined #ocaml
sebz has joined #ocaml
ocp has quit [Read error: Connection reset by peer]
ocp has joined #ocaml
ocp has quit [Client Quit]
ocp has joined #ocaml
gdu has quit [Quit: Quitte]
ocp has quit [Ping timeout: 240 seconds]
ocp has joined #ocaml
edwin has quit [Remote host closed the connection]
eikke has joined #ocaml
ikaros has quit [Quit: Ex-Chat]
ikaros has joined #ocaml
ocp has quit [Quit: Leaving.]
ocp has joined #ocaml
ocp has quit [Ping timeout: 240 seconds]
ocp has joined #ocaml
Morphous has quit [Ping timeout: 240 seconds]
roha has quit [Remote host closed the connection]
ocp has quit [Ping timeout: 240 seconds]
eikke has quit [Ping timeout: 240 seconds]
Xizor has quit []
Morphous has joined #ocaml
sebz has quit [Quit: Computer has gone to sleep.]
eikke has joined #ocaml
pilki has quit [Quit: Leaving]
sebz has joined #ocaml
Cyanure has joined #ocaml
sebz has quit [Quit: Computer has gone to sleep.]
ikaros has quit [Quit: Ex-Chat]
destrius has joined #ocaml
emmanuel_ has joined #ocaml
eikke has quit [Ping timeout: 244 seconds]
Cyanure has quit [Remote host closed the connection]
emmanuel_ has quit [Remote host closed the connection]
emmanuel_ has joined #ocaml
emmanuel_ has quit [Remote host closed the connection]