IRC log of zope3-dev on 2002-03-02

Timestamps are in UTC.

00:04:47 [dreamcatcher]
dreamcatcher has joined #zope3-dev
00:07:20 [dreamcatcher]
cheguei em casa agora
00:11:54 [drauku]
damn my babelfish fell out again
00:13:29 [srichter]
JimFulton: Did you get the main branch working again?
00:17:06 [dreamcatcher]
dreamcatcher has quit
00:20:14 [dreamcatcher]
dreamcatcher has joined #zope3-dev
00:25:04 [dreamcatcher]
anyone can help me to debug a segfault with zope3 on debian-powerpc?
00:26:33 [Metron]
is there a repeatable procedure to check if this segfault is related to the known BTree problem?
00:29:48 [Metron]
guess we could proceed with this question to #zope then :-)
00:30:57 [Martijn]
Martijn is now known as MJ|away
00:36:52 [JimFulton]
Stephan, the tests are still failing
00:37:13 [JimFulton]
Maybe I'll just disable the tests
00:37:43 [srichter]
Jim: Yeah, I will checkout a head too and fix it
00:38:02 [srichter]
Jim: on the other hand I might just merge
00:38:35 [srichter]
Metron: Interesting
00:39:19 [srichter]
Jim: I hope to get XML-RPC working in the next couple hours; geez, it just dawned on me that I need all the ZCML and Service stuff as well...
00:40:12 [dreamcatcher]
srichter: we are trying to backtrace the segfault on debian-powerpc
00:40:30 [dreamcatcher]
srichter: any help would be appreciated ;-)
00:40:57 [srichter]
dreamcatcher: no debian. I have PLENTY of work too
00:41:20 [srichter]
it would be for me I want XUL running NOW. <being stuborn>
00:42:09 [srichter]
guido: I would not have a clue how to fix anything in Python other than in zome high-level Python modules maybe
00:42:51 [guido]
srichter: I don't mean just you, I'm just generally impressed that people here are helping out with Zope3. I wish I knew how to get more folks to help with Python...
00:43:10 [Metron]
guido: Zope3 is written in Python. Python is written in C. Thanks to you, Python rocks. So, basically it's your fault. :-)
00:44:01 [dreamcatcher]
guido: i m reeeeally interested on helping on anything i can!
00:44:52 [dreamcatcher]
guido: but first im needing some help on how to hook gdb to backtrace a segfault
00:45:58 [drauku]
guido, i know C. :) there is a todo list?
00:49:27 [drauku]
wait. i barely have enough time to do zope3
00:51:35 [srichter]
guido: I know you meant this generically, but I wanted to say that it might be easier to help in Zope3 than Python...
00:52:58 [drauku]
debugging a big c program kinda sucks
00:53:51 [limi]
srichter: do you have a few minutes for me?
00:58:02 [srichter]
limi: sure
00:58:52 [srichter]
limi = ? (I am sorry if I forgot and you had told me before)
00:59:18 [limi]
srichter: one of the Plone guys - check your /msg :)
01:01:03 [dreamcatcher]
now i have an output of strace of zope3 on my debian-ppc. i will attach it to the collector entry.
01:01:35 [limi]
srichter?
01:04:40 [JkrLaptop]
JkrLaptop has quit
02:03:52 [blackhole]
blackhole has joined #zope3-dev
02:03:52 [zahn.openprojects.net]
topic is: reminder: the noise level should be kept low. Yes, we know it ain't easy on IRC, but remember, we have #zope and #python.
02:03:52 [zahn.openprojects.net]
Users on #zope3-dev: blackhole hazmat Metron limi @guido rdmurray @JimFulton pobrien tav HammerToe Ipazuzu scipient SteveA|way srichter MJ|away FarcePest cdesousa bigdaddy @drauku
02:14:17 [zigg]
zigg has joined #zope3-dev
02:14:19 [limi]
limi has quit
02:25:50 [JimFulton]
JimFulton has quit
02:28:01 [limi]
limi has joined #zope3-dev
02:28:33 [zigg]
hmm. OOBTree is definitely giving my gc headaches. I want to know why it gives nobody else headaches. can I get someone to try something for me?
02:39:40 [srichter]
zigg: shoot
02:39:55 [zigg]
PYTHONPATH=lib/python python
02:39:58 [zigg]
import gc
02:40:04 [zigg]
import Persistence.BTrees.OOBTree
02:40:06 [zigg]
gc.collect()
02:40:09 [zigg]
see if that crashes your python
02:40:27 [srichter]
ok
02:40:50 [zigg]
if it doesn't, I'm going to scream like a little girl. (-:
02:41:39 [drauku]
record it
02:41:49 [srichter]
I got 0 as result
02:41:53 [srichter]
no crash
02:41:56 [zigg]
grr.
02:42:10 [zigg]
well, that rules out yet another idea
02:42:12 [srichter]
that is a build with the OOTree stuff from last night
02:42:21 [zigg]
I think I updated two days ago, I'll update again just in case.
02:42:28 [srichter]
on: [GCC 2.96 20000731 (Red Hat Linux 7.1 2.96-81)] on linux2
02:42:59 [zigg]
drau: ?
02:43:30 [zigg]
ah, more changes :-)
02:43:43 [zigg]
let's see...
02:43:58 [srichter]
zigg: let me know...
02:45:37 [zigg]
urg. now I can't even get to gc.collect()
02:52:53 [srichter]
zigg: what error do you get?
02:53:25 [zigg]
core
02:53:38 [zigg]
when I look at it in gdb, the garbage collector structure seems to be unitialized.
02:53:40 [srichter]
core dump? Shrug
02:53:44 [zigg]
:-)
02:54:00 [srichter]
weired
02:54:23 [srichter]
could this be a memory management problem?
02:54:43 [zigg]
I'm really not sure, I'm not familiar with python internals at all.
02:54:44 [drauku]
zigg this is on openbsd?
02:54:46 [zigg]
yes
02:54:55 [drauku]
srichter, is on linux or windows
02:55:03 [drauku]
s/,//
02:55:24 [drauku]
if all things being equal, python is borked maybe for openbsd?
02:55:37 [zigg]
well, python 2.1.2 works just great :-)
02:55:38 [zigg]
brb
02:55:57 [srichter]
drauku: linux, same box you work on
02:56:41 [drauku]
ok
02:56:57 [drauku]
zigg, okie dokie, sorry for the "insult" there then :)
02:58:15 [srichter]
zigg: you have a self built Python, right?
03:05:57 [Metron]
Metron has quit
03:09:06 [zigg]
back
03:09:24 [zigg]
I'm running the entirely 'default' install.
03:09:32 [zigg]
I suppose I could go play with some compiler flags...
03:10:12 [zigg]
there is no Python binary for OpenBSD except the port I maintain, and that one is dying too, except I can't get a backtrace because it has no debug symbols.
03:11:31 [srichter]
zigg: :(
03:12:05 [zigg]
it will work if I --without-cycle-gc it :-)
03:24:04 [srichter]
what does that mean?
03:24:57 [zigg]
if I compile python turning off the cycle garbage collector
03:25:09 [zigg]
it basically means anything that can't be picked up by the reference counter will never be garbage collected
03:25:26 [zigg]
sure, it'll work great for development, but it'd be nice to get it fixed earlier rather than later...
03:28:34 [pobrien]
pobrien has quit
03:34:54 [dreamcatcher]
dreamcatcher has joined #zope3-dev
03:37:01 [srichter]
zigg: right
03:39:17 [zahn.openprojects.net]
Disconnected from zahn.openprojects.net (Connection reset by peer)
04:06:00 [blackhole]
blackhole has joined #zope3-dev
04:06:00 [zahn.openprojects.net]
topic is: reminder: the noise level should be kept low. Yes, we know it ain't easy on IRC, but remember, we have #zope and #python.
04:06:00 [zahn.openprojects.net]
Users on #zope3-dev: blackhole bigdaddy cdesousa MJ|away srichter SteveA|way Ipazuzu HammerToe tav rdmurray limi scipient zigg hazmat @drauku
04:12:50 [YET_____]
YET_____ has joined #zope3-dev
04:13:13 [srichter]
srichter has quit
04:16:06 [JkrLaptop]
JkrLaptop has joined #zope3-dev
04:23:47 [djay]
djay has joined #zope3-dev
04:33:29 [zigg]
zigg has quit
05:42:58 [blackhole]
blackhole has joined #zope3-dev
05:42:58 [zahn.openprojects.net]
topic is: reminder: the noise level should be kept low. Yes, we know it ain't easy on IRC, but remember, we have #zope and #python.
05:42:58 [zahn.openprojects.net]
Users on #zope3-dev: blackhole dc0e troy_es djay YET_____ bigdaddy cdesousa MJ|away SteveA|way Ipazuzu HammerToe tav rdmurray limi scipient @hazmat @drauku
08:46:44 [blackhole]
blackhole has joined #zope3-dev
08:46:44 [zahn.openprojects.net]
topic is: reminder: the noise level should be kept low. Yes, we know it ain't easy on IRC, but remember, we have #zope and #python.
08:46:44 [zahn.openprojects.net]
Users on #zope3-dev: blackhole mithro dc0e YET_____ bigdaddy cdesousa MJ|away SteveA|way Ipazuzu HammerToe tav rdmurray @hazmat
19:05:15 [blackhole]
blackhole has joined #zope3-dev
19:05:15 [zahn.openprojects.net]
topic is: ::: [ z o p e 3 - d e v ] ::: reminder: the noise level should be kept low. ::: if you have logs from the last two days, please email them to me [[email protected]], so that they can be put up on the web logs of the channel. thanks :::
19:05:15 [zahn.openprojects.net]
Users on #zope3-dev: blackhole guido limi acamargo srichter zigg JkrLaptop xena Ipazuzu tav drauku dc0e YET bigdaddy cdesousa MJ|away SteveA HammerToe rdmurray @hazmat
19:09:59 [hazmat]
my logger keeps timing out :(
19:10:21 [zigg]
xena is here now ...
19:10:27 [hazmat]
cool
19:10:41 [hazmat]
zigg: about the wrapper, did you run stupid_build.py
19:10:42 [SteveA]
I have logs since yesterday morning
19:10:51 [zigg]
haz: yes, I realized that a few seconds later ;-)
19:10:57 [zigg]
I mean, yes, I forgot to run it
19:11:18 [hazmat]
i've got an automated setup to upload logs to zope.org, except the logger isn't staying connected reliably.
19:13:01 [zigg]
I need to get in touch with some python internals people about this garbage collector issue... something changed between 2.1 and 2.2 that is giving OpenBSD fits.
19:13:39 [zigg]
anyone got any suggestions?
19:13:53 [zigg]
for people I mean, I don't want to waste anymore zope3-dev time with this :-)
19:20:57 [SteveA]
SteveA is now known as SteveA|way
19:24:27 [srichter]
did you know? Traversers are only "Views" in Zope 3
19:24:30 [srichter]
interesting
19:32:29 [Metron]
Metron has joined #zope3-dev
19:33:33 [Metron]
Metron has quit
19:33:54 [Metron]
Metron has joined #zope3-dev
19:34:35 [dreamcatcher]
dreamcatcher has joined #zope3-dev
22:34:32 [blackhole]
blackhole has joined #zope3-dev
22:34:32 [zahn.openprojects.net]
topic is: ::: [ z o p e 3 - d e v ] ::: reminder: the noise level should be kept low. ::: if you have logs from the last two days, please email them to me [[email protected]], so that they can be put up on the web logs of the channel. thanks :::
22:34:32 [zahn.openprojects.net]
Users on #zope3-dev: blackhole strichter @lalo[out] guido zigg JkrLaptop xena Ipazuzu tav drauku dc0e bigdaddy cdesousa MJ|away SteveA|way HammerToe rdmurray @hazmat
22:38:44 [guido]
blackhole: who are you
22:38:44 [guido]
I'm logging. I don't understand 'who are you', guido. Try /msg blackhole help
22:39:33 [zigg]
blackhole is hazmat's bot
22:39:44 [guido]
zigg: I know, just playing with him :)
22:39:47 [zigg]
o :-)
22:39:51 [hazmat]
which sadly disconnects at random...
22:41:57 [strichter]
does anyone know where the code for the tab creation is hidden/
22:41:59 [strichter]
?
22:43:58 [hazmat]
for the zmi?
22:44:03 [hazmat]
its in ZMIViewService
22:44:23 [strichter]
yes
22:44:44 [hazmat]
tabs are registered in config files, the ZMIViewService makes them available and filters them according to security and optionally a TALES expression.
22:45:16 [hazmat]
although how i implemented the security check should probably be revisited...
22:45:21 [strichter]
so, all I got to say is ZMIViewService.getViews() ?
22:46:18 [hazmat]
needs a context, i believe... i'm a little rusty.
22:46:28 [strichter]
yeah, I just saw that
22:46:37 [strichter]
What does it return?
22:46:45 [strichter]
list of ...?
22:46:53 [zigg]
whoa... ok... I am no C genius by any stretch, but I think there is a major GC-related data corruption problem with Python 2.2
22:47:37 [guido]
zigg: what makes you think that? I'm not aware of any.
22:47:58 [zigg]
well, PyType_Type has the HAVE_GC flag set.
22:48:02 [zigg]
or HAS_GC, I forget right now :-)
22:48:20 [guido]
...and?
22:48:21 [zigg]
all the type initializations I've seen are static
22:48:26 [zigg]
with no room for the GC head structure
22:48:31 [hazmat]
strichter: list of tuple corresponding to label and paths
22:48:54 [guido]
zigg: we've taken care of that, hold on why I get the reference to look up...
22:49:04 [zigg]
hmm, ok
22:49:24 [guido]
zigg: check out type_is_gc() in typeobject.c.
22:49:29 [strichter]
hazmat: :( paths are view-type specific
22:49:37 [hazmat]
perhaps there should be a #python-dev channel.
22:49:53 [guido]
hazmat: but I normally don't have time for that
22:49:57 [hazmat]
ic
22:50:01 [zigg]
haz: probably... but I'm still trying to determine whether my issues are z3-specific or python-specific :-)
22:50:49 [hazmat]
strichter: how so, the paths don't have to correlate to actual posititon if the default traverse chooses to return a different path segment upon traversal to the path.
22:51:51 [strichter]
mmh, ok, so my view object would need the same name for XUL and Browser... that's ok
22:52:19 [hazmat]
perhaps the ZMIviewservice, should taken into account various presentation types, and maintain registries if the existing isn't flexible enough.
22:52:31 [strichter]
yep
22:52:42 [strichter]
I think that might be necessary
22:52:46 [strichter]
later
22:52:51 [strichter]
for now it is ok
22:58:50 [strichter]
hazmat: actually I realized that the thing with the tabs is not okay, since both HTML and XUL use the Browser View namespace
22:59:31 [zigg]
woo! fixed!
23:00:56 [strichter]
zigg: VERY GOOD JOB!
23:01:12 [zigg]
if it weren't for guido's hint, I'd still be running around in circles. :-)
23:01:25 [strichter]
zigg: If everyone cares about porting that much, we will have the most portable platform ever! :)
23:01:45 [strichter]
guido: good job in giving hints to zigg. :)
23:01:56 [guido]
I have no idea what I did :)
23:02:01 [strichter]
LOL
23:02:01 [zigg]
heheh
23:02:02 [lalo[out]]
whoa
23:02:22 [lalo[out]]
tried to test the Formulator branch, got three pages worth of traceback
23:02:25 [zigg]
well, I'm just driven by an intense desire to not reformat my laptop.
23:02:28 [strichter]
zigg: but now you need to let us know...
23:02:37 [zigg]
yes, well I am going to take this one to jeremy...
23:02:47 [strichter]
lalo: can you send them to me...
23:03:05 [zigg]
brb
23:03:15 [strichter]
lalo: unfortunately my cvs access is broke, but I might be able to give hints
23:03:23 [lalo[out]]
I'll try to re-run stupid_build first to be sure it isn't my fault
23:03:30 [strichter]
zigg is getting a bottle of champagne to celebrate
23:03:57 [strichter]
lalo: ok. I had only three errors
23:04:06 [strichter]
lalo: which is ok
23:04:44 [lalo[out]]
I can't dcc
23:04:48 [lalo[out]]
(firewall)
23:05:04 [lalo[out]]
but seems to be something related to site.zcml
23:05:12 [strichter]
?
23:05:17 [zigg]
strichter: already had a couple beers, I'm set. and I relearned gdb, which in itself kills more brain cells than most alcoholic celebrations. and now I'll stop lowering the S/N ratio of the channel (-:
23:05:32 [strichter]
lalo: send site.zcml and errors to [email protected]
23:05:52 [lalo[out]]
nope; actually it's about App/OFS/Folder/Views/Browser/browser.zcml (nice that zcml traces back too)
23:06:10 [strichter]
lalo: yep.
23:06:15 [rdmurray]
rdmurray has quit
23:07:35 [lalo[out]]
done
23:08:10 [lalo[out]]
and I'm gone :-) bbl
23:09:02 [strichter]
lalo: bye
23:21:57 [strichter]
hazmat: do you remember the ZPT file where you generate the Tabs part?
23:23:49 [SteveA|way]
SteveA|way is now known as SteveA
23:23:55 [SteveA]
I know about ZMI tabs
23:24:15 [strichter]
SteveA: Hello, yes, you know where the ZPT is?
23:24:27 [SteveA]
just a sec
23:24:36 [strichter]
SteveA: BTW, is your CVS access also shot?
23:25:03 [SteveA]
just updated ok
23:25:12 [strichter]
:(
23:25:16 [SteveA]
Zope/App/ZMIViewUtility.py
23:25:53 [SteveA]
had to look into this when I was learning about the ContextWrapper thing
23:26:04 [hazmat]
standard_macros.pt
23:26:07 [strichter]
ah
23:26:12 [hazmat]
in Zope/App/ZMI/www
23:26:33 [SteveA]
oh yeah... I answerwed a similar but related question :-)
23:26:36 [strichter]
ah, ther it is
23:27:04 [hazmat]
hi SteveA
23:27:31 [SteveA]
hi K.
23:28:32 [SteveA]
cool.
23:28:44 [SteveA]
a "Sprint for Zope3" one?
23:29:43 [hazmat]
no a thanks for talking about zope at ipc10, type of t-shirt, its a standard zope.org t-shirt with a zope-corp label embrodiered (sp:( ) on the front
23:31:14 [SteveA]
oh, cool
23:31:21 [hazmat]
i now have enough python and zope tshirts for a week and a half :)
23:31:33 [SteveA]
I wonder if I can blag one for talking about Zope3 at the uk python conference next month :)
23:31:37 [strichter]
look at: http://physics.cbu.edu:8080/loaded/main_xul;view
23:32:07 [strichter]
with NS 6.2 or moz of course
23:32:28 [hazmat]
cool
23:32:32 [zigg]
that's another thing I have to do, get mozilla running on OpenBSD (_:
23:32:46 [hazmat]
the menu on the left looks a little funny at the bottom left though... not sure why.
23:33:35 [strichter]
this is because I stretched the box only for 600 px. vertically
23:33:44 [strichter]
height="100%" does not work
23:33:57 [zigg]
ick.
23:34:13 [zigg]
I'm of the firm opinion that pixel values should not go in the tree... for what it's worth.
23:34:16 [SteveA]
strichter: cool. Can you do them as tabs too? :)
23:34:31 [strichter]
SteveA: Yes
23:35:03 [zigg]
s: did you submit a moz bug about that?
23:35:04 [SteveA]
I'm about 3/4 through the XULPlanet tutorial
23:35:05 [strichter]
but I thought this way it would be better, since we had this tab convolution in Z2
23:35:20 [strichter]
yeah, the tutorial is quiet nice
23:35:58 [strichter]
zigg: I agree. In the end there will be no pixels left
23:36:10 [zigg]
cool. 10x. :-)
23:36:20 [hazmat]
the XULTutorial is great, another good bit of reading is the oreilly javscript book 3rd edition
23:36:25 [hazmat]
for moz programming.
23:36:53 [strichter]
I have only one big problem let
23:37:23 [strichter]
I cannot get around the security that will not permit Web-based scripts to access XPComponents, such as XML-RPC
23:37:35 [strichter]
BTW, I have a half-baked version of XML-RPC running too
23:38:03 [hazmat]
you have to run and install it as a xul app, most likely. you should have access to XmlHttpRequest though.
23:38:12 [strichter]
But Zope/App/ZopePublication needs a MAJOR redesign in order to allow for other protocols
23:38:45 [hazmat]
indeed. how are you deciding between protocols layered over top of http (ie xml-rpc vs. soap)?
23:38:53 [strichter]
hazmat: But Martijn worte such a nice XML-RPC support for Mozilla... why would I want to go that low level
23:39:24 [strichter]
hazmat: Well, that is easy, since you do that with Payloads...I can do that right now
23:39:26 [hazmat]
strichter: there probably is a way, but i'm fairly clueless on moz security architecture.
23:39:37 [strichter]
but the actual Traveral is totally browser-specific
23:39:55 [strichter]
in fact, even the naming convention in IBrowserPublisher sucks
23:40:02 [strichter]
it should be IHTTPPublisher
23:40:24 [strichter]
then I can reuse that stuff for Browser, XML-RPC and WebDAV
23:41:50 [strichter]
In the Traverser it says: IBrowserPublisher.isImplementedBy(ob) ... If this would be avstracted, I could reuse the Traverser code, but this way I can't
23:42:12 [lalo[out]]
lalo[out] has quit
23:42:19 [strichter]
but back to the XUL security
23:42:51 [strichter]
But I think I do not want to install a package
23:43:09 [zigg]
isn't there code-signing ability, much like there is for Java?
23:43:22 [SteveA]
can you get it so that you can call XML-RPC on the site that served you the page?
23:43:26 [strichter]
I want to set security in a way that the user can say: "I am aware of a possible security issue, but still allow it for this domain"
23:43:27 [zigg]
not that that's a particularly good way to do it :-)
23:43:28 [SteveA]
like with Java applets
23:43:36 [strichter]
zigg: yes
23:43:56 [zigg]
I believe that with Java applets, you have to sign the applet before you can get to the "please let my applet do this" prompt
23:44:11 [strichter]
that's exactly what I want too
23:44:22 [strichter]
I think you can do this with signed scripts
23:44:52 [SteveA]
<strichter> that's exactly what I want too <-- was that in answer to me?
23:44:54 [zigg]
well, if you determine that that's what you need to do, let me know, and I can fudge up a fake CA for testing.
23:45:06 [strichter]
SteveA: yes
23:45:21 [strichter]
CA?
23:45:42 [strichter]
I think I need to talk to Martijn about this
23:45:46 [zigg]
probably :-)
23:45:50 [strichter]
he should know much more...
23:46:35 [zigg]
I think the old zope-mozilla stuff required some things to be installed in mozilla's chrome dir.
23:46:38 [strichter]
I found on Web pages, that you can actually set Mozilla prefs to get around the security (at least for development purposes), but none of the settings worked
23:47:06 [strichter]
zigg: Well, Zope Studio tried to be a real application on the server side
23:47:25 [strichter]
but I want to avoid that, so that we are more flexible to changes in Zope
23:47:25 [SteveA]
strichter: in that case, if xmlrpc in mozilla doesn't work that way now, it is a reasonable enhancement to request
23:47:29 [hazmat]
a signed script wouldn't do it, since all the pages would need to be signed.
23:47:45 [strichter]
hazmat: yep, I know :(
23:48:04 [hazmat]
i guess you could get by if you restrict a transport layer to a signed script.... not sure.
23:48:22 [strichter]
mmh, interesting
23:48:35 [strichter]
but all this still sounds like a big hack
23:48:56 [strichter]
I do not understand what the problem with running XML-RPC client side is
23:49:07 [strichter]
it si the same as calling a Web site...geez
23:49:54 [SteveA]
anyone got a bugzilla.mozilla.org account?
23:50:55 [strichter]
nope
23:51:03 [strichter]
I guess I could get one
23:51:44 [strichter]
but I think we will get a standard answer that XML-RPC is a XPComponent and therefore follows this specific security rules
23:55:37 [strichter]
I will wait until I get a hold of MArtijn
23:57:12 [zigg]
I have a bugzilla account.
23:57:12 [SteveA]
SteveA is now known as SteveA|way