| # | Time | Nick | Message |
|---|
| # | 07:48:13 | sfortin has joined #evergreen |
| # | 07:54:33 | mck9 has joined #evergreen |
| # | 08:10:46 | ToPdOg has quit IRC |
| # | 08:25:43 | Dyrcona has joined #Evergreen |
| # | 08:38:09 | Dyrcona | Yesterday I proposed using libcrypto's digest implementation for sha1 in libopensrf on the mailing list. |
| # | 08:38:43 | Dyrcona | Today, I have a patch against trunk, but I have a question about submitting it because it adds two new files: digest.h and digest.c. |
| # | 08:39:02 | Dyrcona | I'm not sure how I should add the new files in the patch/attachment. |
| # | 08:44:39 | berick | Dyrcona: I don't think it's unreasonable to send the patch plus any new files as their own attachments w/ some description of where they need to go |
| # | 08:45:25 | Dyrcona | ok. should i just send them to the dev list or does anyone else want to see them first? |
| # | 08:45:52 | berick | dev list is best |
| # | 08:46:13 | Dyrcona | I should probably try it on our opensrf server, first. |
| # | 08:46:22 | berick | that would be good :) |
| # | 08:46:31 | Dyrcona | i've made sure that it compiles, but I've not run it, yet. |
| # | 08:46:40 | berick | unless your just submitting it as a proof of concept.. right |
| # | 08:47:08 | berick | Dyrcona++ |
| # | 08:47:23 | Dyrcona | could be proof of concept, but I think overall using libcrypto is better than having our own implementations of sha1 and md5. |
| # | 08:47:53 | berick agrees |
| # | 08:48:33 | Dyrcona | I'm out of the office today, so it could be tomorrow or early next week before I can submit the patch. |
| # | 08:51:25 | berick | mck9: ^-- |
| # | 08:52:38 | bshum has joined #evergreen |
| # | 08:52:50 | mck9 | berick: are you pinging me? |
| # | 08:53:00 | bshum has quit IRC |
| # | 08:53:13 | bshum has joined #evergreen |
| # | 08:53:14 | berick | mck9: just making sure you saw the converstaion above, since you were looking into the mde5/sha code |
| # | 08:54:25 | mck9 | Yeah, well it may be a while before I do much with it. For the foreseeable future we can expect to keep the existing sha1 code, unless you want me to escalate it. |
| # | 08:55:21 | mck9 | Now if the proposed patch is a replacement of the sha1 code, then that's another matter. |
| # | 08:56:40 | berick | mck9: i think that (patch == replacemnet) is a distinct possibility |
| # | 08:56:55 | jeff | heh. i just responded to that thread on -dev |
| # | 08:59:22 | mck9 | I didn't know i was stirring up a hornet's nest. Thanks jeff -- incompatibility with the GPL is a deal-breaker. |
| # | 08:59:27 | jeff | OpenSSL apparently has GPL incompatabilities. The OpenSSL FAQ suggests that if you are developing open source software that uses OpenSSL, you consider licensing your software under a license other than the GPL. |
| # | 09:00:00 | berick | wow |
| # | 09:00:13 | jeff | (they also suggest as an alternative adding an exemption clause to your GPL license statement) |
| # | 09:00:26 | jeff | and "If you are using GPL software developed by others, you may want to ask the copyright holder for permission to use their software with OpenSSL." |
| # | 09:00:38 | jeff | This is how things like gnutls came about. |
| # | 09:00:53 | jeff | http://www.openssl.org/support/faq.html#LEGAL2 |
| # | 09:00:53 | jeff | http://www.gnome.org/~markmc/openssl-and-the-gpl.html |
| # | 09:01:05 | jeff | for background/details if anyone cares. |
| # | 09:11:24 | mck9 | From looking at jeff's links, it might be possible to squeeze through a legal chink and use OpenSSL, but it's a very dodgy thing. I don't think we should touch it if there's any way to avoid it. |
| # | 09:12:57 | phasefx_ | KISS |
| # | 09:14:23 | Dyrcona | I don't see it being a problem as long as opensrf does not include openssl itself. |
| # | 09:14:39 | Dyrcona | From the openssl FAQ: On many systems including the major Linux and BSD distributions, yes (the GPL does not place restrictions on using libraries that are part of the normal operating system distribution). |
| # | 09:15:30 | mck9 | Then we would restrict EG to systems that already provide the OpenSSL library. |
| # | 09:16:05 | Dyrcona | OpenSSL has to be there already or most of the components don't work anyway. |
| # | 09:17:29 | Dyrcona | After all, you're already expecting Apache to provide SSL. Postrgres practically requires SSL for remote connections. Several jabber servers require it, too. |
| # | 09:18:38 | berick | of course, we can also consider the fact that everything is (technically) working just fine right now ;) |
| # | 09:18:42 | berick devils advocate |
| # | 09:19:17 | mck9 | Also the "special exception" in the GPL may be only in version 3. If so, then we would have to make EG a GPL v3-only project, instead of v2-or-v3-at-your-option. |
| # | 09:19:17 | jeff | berick: surgeons general? ;-) |
| # | 09:20:01 | jeff | mck9: the exemption in question is present in GPLv2 |
| # | 09:20:19 | mck9 | Okay thanks, I hadn't checked yet |
| # | 09:20:50 | mck9 | Other components may require SSL, but that doesn't mean they require OpenSSL specifically. |
| # | 09:20:54 | jeff | either change introduces an additional dependency, but reduces reinventing the wheel. |
| # | 09:21:31 | jeff | in the case of openssl, it's another dev package to install (i don't think anything needs the dev package yet), and a potential licensing gotcha. |
| # | 09:21:43 | jeff | in the case of mhash, it's another dev package and it's LGPL'd |
| # | 09:21:54 | jeff | i don't know how easy the code against either is. |
| # | 09:23:23 | phasefx_ | for Debian, we'd have to include a GPL exception if we link against openssl |
| # | 09:23:47 | mck9 | If mhash is LGPL we can include its source, or a subset thereof, in our own distribution, the way we currently include the sha.c file. No need to load a dev package. |
| # | 09:23:57 | jeff | phasefx_: oh? i didn't catch that part. |
| # | 09:24:58 | jeff | mck9: yes, an option we wouldn't have with openssl, unless we complied with both of the advertising clauses. |
| # | 09:25:37 | mck9 | It's mainly the advertising clauses that create the problem, apparently. |
| # | 09:25:53 | Dyrcona | yep. stupid advertising clauses.... |
| # | 09:25:58 | jeff | if the dev libs in question are already packaged for all of the distributions that evergreen is supported on, it might be more maintainable in the long run to use those packages. depends on how stable things are. |
| # | 09:26:18 | phasefx_ | http://bugs.debian.org/cgi-bin/bugreport.cgi?bug=501145 http://lists.debian.org/debian-legal/2002/10/msg00113.html |
| # | 09:26:31 | mck9 | The GPL says we can't require downstream recipients to comply with an advertising clause, as I understand it. |
| # | 09:27:00 | Dyrcona | Hrm, should I just rework the code to use GNUTLS, then? It should be available on the distros we're aiming at? |
| # | 09:27:34 | phasefx_ | is libcrypto part of openssl? |
| # | 09:27:46 | Dyrcona | Yes, I believe so. |
| # | 09:27:55 | phasefx_ | ah, soo desu |
| # | 09:29:27 | Dyrcona is not a big fan of copying implementation code from other projects, particularly libraries. |
| # | 09:29:49 | Dyrcona prefers linking against libraries so he has less implementation to worry about. :) |
| # | 09:30:27 | dbs has joined #evergreen |
| # | 09:30:49 | phasefx_ | dan the man |
| # | 09:30:53 | phasefx_ bets he hates being called that |
| # | 09:30:58 | mck9 | Drcona: Agreed. But copying is an option for some and not for others, and I like to keep my options open. |
| # | 09:31:02 | dbs doesn't mind that much |
| # | 09:31:38 | dbs | "dan the man who screwed things up" on the other hand... truth always hurts |
| # | 09:32:46 | phasefx_ | dan with the plan |
| # | 09:32:47 | dbs | jeff: licensing concerns? whatever... (apparently that's the thing to say) |
| # | 09:33:25 | jeff | phasefx_: wait, dbs is a cylon? |
| # | 09:33:32 | phasefx_ | supposed to say "whatever" with your inside voice |
| # | 09:33:42 | dbs | heh |
| # | 09:33:54 | jeff | BY YOUR WHATEVER. |
| # | 09:34:52 | mck9 | dbs: just before you came in we were discussing licensing issues with OpenSSL. |
| # | 09:35:21 | dbs reads scrollback, favours package linking (or compiling the source package if a package isn't available *ahem* libmemcached *ahem*) |
| # | 09:36:16 | dbs | otherwise, I haven't coded against either mhash or libcrypto so have no opinion on the ease of use of either one |
| # | 09:37:16 | dbs | jeff++ # by your SUDO |
| # | 09:41:33 | jeff | alias whatever,=sudo |
| # | 09:42:31 | Dyrcona | mhash looks like the way to go in light of the libcrypto license issue. |
| # | 09:45:18 | dbs was peeking around at heimdal just to throw another option on the table, but it might be more complex |
| # | 09:46:00 | gmcharlt | miker_++ # twisting tongues with the best of them |
| # | 09:48:14 | dbs | http://www.h5l.org/manual/HEAD/hcrypto/group__hcrypto__evp.html |
| # | 09:52:07 | alxp has joined #evergreen |
| # | 09:52:08 | Dyrcona | That looks just like the OpenSSL interface. |
| # | 09:52:15 | moodaepo has joined #evergreen |
| # | 09:53:19 | phasefx_ | GNU TLS has an OpenSSL compatibility layer, fwiw |
| # | 09:56:23 | dbs | Dyrcona: ah, that's nice; by complex I meant trying to work out which debian package (if any) one would be able to depend on |
| # | 09:59:09 | Dyrcona | I'm going to install gnutls on my laptop and try my code with it.--I did check that not only my function compiles, but it gives a desirable result. |
| # | 10:00:55 | dbs | phasefx: Dyrcona: IIRC, packages with sophisticated autotools setups can default to linking against gnutls but fall back to libopenssl |
| # | 10:01:33 | dbs | same with readline vs libedit and other GPL-compatible vs. GPL-incompatible libraries |
| # | 10:02:44 | phasefx_ was wondering about that, "how sophisticated are we? :) |
| # | 10:04:05 | mck9 | FWIW I'm writing a summary of this discussion and will be sending it to the dev list once the discussion settles down, or I get tired of it. Some people may be on holiday and would otherwise miss it. |
| # | 10:07:09 | djfiander has joined #Evergreen |
| # | 10:07:38 | dbs | an example of readline vs. libedit (the example I was thinking of, actually): http://svn.php.net/viewvc/php/php-src/trunk/ext/readline/config.m4?revision=292081&view=markup |
| # | 10:07:42 | dbs | phasefx: not very :) |
| # | 10:09:37 | dbs | djfiander: SIP was holding up reasonably well for a few tests yesterday, shock and surprise, although the 3M guy told me that SIP2 over raw sockets doesn't use username/password on "the mythical beast" systems, which sounds insane |
| # | 10:09:50 | dbs | djfiander++ |
| # | 10:10:07 | djfiander | that is whacked. |
| # | 10:10:42 | djfiander | one can only hope that one has to register the IPs of the clients with the ILS, or something |
| # | 10:10:46 | dbs | he actually had me set up the system and try connecting without a username / password to begin with |
| # | 10:11:59 | dbs | no hope here, only despair :) |
| # | 10:12:55 | dbs | mck9++ # sounds like a good idea |
| # | 10:13:13 | djfiander | sometimes the appropriate response to being told to "be generous in what you accept" is, "Fuck no. get the protocol right, asshole." |
| # | 10:15:51 | dbs | my despair was doubled by the tech guy strongly suggesting that we keep the administrative passwords on the self-check machine the defaults |
| # | 10:16:06 | atz | dbs: that's why it's raw |
| # | 10:16:14 | mck9 | djfiander: Another way to translate "be generous in what you accept" is "change the rules arbitrarily for your own convenience." |
| # | 10:16:48 | dbs | and then redoubled by our computer services person suggesting that we create a shared VPN account for which all of the circ staff would have the username / password |
| # | 10:17:40 | atz | dbs: welcome to a technology stuck in the era of serial port connectivity. |
| # | 10:18:17 | djfiander | dbs: for the admin passwords on the self-check machines. They do have the advantage that they are locked in a cabinet, and in most libraries would not be managed by IT staff but by branch staff. So, depending on physical security isn't a big issue. |
| # | 10:18:23 | dbs | atz: seriously? it's bad enough that the telnet option passes cleartext |
| # | 10:18:50 | djfiander | dbs: the protocol was designed for hardline connections. security isn't a big deal ;-) |
| # | 10:19:17 | dbs | (yes, I know that's the definition of telnet) |
| # | 10:20:19 | atz | dbs: yeah, the 3M emulator/test apparatus was written in 1997 on top of NT4 and a foxpro database. |
| # | 10:20:32 | dbs | anyhoo, raw SIP2 over port-forwarded connection is working nicely, if a tad slow |
| # | 10:21:06 | djfiander | fortunately self-check is neither particularly time sensitive nor a heavy network load. |
| # | 10:25:36 | dbs | atz: weird. IIRC, the internet existed in 1997. oh well, I'll shut up. |
| # | 10:28:17 | Dyrcona | dbs: For most commercial ILS vendors, the internet still doesn't exist.... |
| # | 10:28:41 | godzilla8nj has joined #evergreen |
| # | 10:29:17 | Dyrcona | Well, GNUTLS is not working for me. The OpenSSL compatibility lacks a few things that I used from OpenSSL. |
| # | 10:29:27 | dbs heads off to test & document the self-check machine for staff |
| # | 10:30:04 | godzilla8nj | can anyone please help: i've been struggling for days to get opensrf 1.2 installed. have finally traced it back to ejabberd failing, but i have no idea how to fix it. |
| # | 10:34:36 | atz | got error msgs? |
| # | 10:35:59 | godzilla8nj | E(<0.314.0>:ejabberd_listener:106) : Failed to open socket for {7777... |
| # | 10:36:35 | godzilla8nj | (from /var/log/ejabberd/ejabberd.log) |
| # | 10:43:13 | godzilla8nj | without a clue as to what i'm talking about, i think it has to do with the erlang node name. per step #5 of http://www.open-ils.org/dokuwiki/doku.php?id=opensrf:1.2:install i set it to ejabberd@localhost but in /var/log/ejabberd/ejabberd.log i see an early error that says it's set to ejabberd@(mydomain.com) |
| # | 10:48:17 | moodaepo | godzilla8nj: What do you see in /etc/default/ejabberd? |
| # | 10:50:15 | dbs has quit IRC |
| # | 10:50:52 | lisppaste3 | godzilla8nj pasted "/etc/default/ejabberd" at http://paste.lisp.org/display/92534 |
| # | 10:51:50 | sfortin has quit IRC |
| # | 10:56:03 | lisppaste3 | moodaepo annotated #92534 "my /etc/default/ejabberd" at http://paste.lisp.org/display/92534#1 |
| # | 10:57:09 | lisppaste3 | moodaepo pasted "ejabberd install info" at http://paste.lisp.org/display/92535 |
| # | 11:00:27 | dbs has joined #evergreen |
| # | 11:03:53 | godzilla8nj | is there a command that erlang requires in order to accept the adjusted node name from /etc/default/ejabberd? |
| # | 11:24:30 | sfortin has joined #evergreen |
| # | 11:42:16 | dbs | djfiander: hey... getting a checksum error on patron summaries or checkouts for titles containing accents. counting bytes instead of characters or something? |
| # | 11:42:30 | djfiander | hmm |
| # | 11:42:43 | dbs | same problem whether I tell the self-check to use ANSEL, Latin-1, or UTF-8 |
| # | 11:43:02 | djfiander | that came up back in the PINES pre-launch days with a spanish Harry Potter book. There's a unit test for it. |
| # | 11:43:05 | dbs | (I assumed that switching from latin-1 to UTF8 would fix the problem, but no...) |
| # | 11:43:23 | dbs | fortunately there's almost no french content here! |
| # | 11:43:32 | djfiander | odd |
| # | 11:43:44 | djfiander | can you build a test case and send it to me? |
| # | 11:43:59 | djfiander | I'll look at it over the break. |
| # | 11:44:08 | djfiander | (and bill Equinox) ;-) |
| # | 11:44:15 | dbs | djfiander: I will try |
| # | 11:44:29 | dbs | merci, djfiander |
| # | 11:44:49 | djfiander | *think* |
| # | 11:45:58 | djfiander | send me the error log message too, if you can. |
| # | 11:46:05 | dbs | I'll try to dig into the SIP2 code a bit further, it will probably benefit me in the long run |
| # | 11:46:45 | dbs | sure, I've got the logs running at debug level on the self-check unit (which is how I can see it's a checksum error) and will try to sync that up with our EG logs |
| # | 11:48:40 | brendan_ga has quit IRC |
| # | 11:49:43 | dbs has quit IRC |
| # | 12:03:35 | godzilla8nj has quit IRC |
| # | 12:14:57 | dbs has joined #evergreen |
| # | 12:15:53 | Dyrcona | I have a working char *shahash(const char *str) using gnutls's sha1 implementation. |
| # | 12:16:05 | Dyrcona | it gives the same results that I got using OpenSSL. |
| # | 12:16:16 | dbs | Hmm. vim, at least, says that the encoding of SIPtest.pm is latin-1 |
| # | 12:16:34 | dbs | but I've gotta go and bake a carrot cake for A the younger's birthday |
| # | 12:16:39 | dbs | Dyrcona++ |
| # | 12:16:54 | berick | dbs: birthday is today? |
| # | 12:16:57 | Dyrcona | I'll send a patch to the list tomorrow after I get a chance to test it on our server. |
| # | 12:17:00 | dbs | tomorrow |
| # | 12:17:01 | berick | Dyrcona++ |
| # | 12:17:20 | dbs | family is here to celebrate tonight, though, so we go with what we've got :) |
| # | 12:17:47 | dbs | he had 15 minutes to spare before he would have been obligated to be named "jesus" |
| # | 12:17:51 | djfiander | dbs check the encoding on the test data file too rmemeber. |
| # | 12:18:21 | berick | dbs: well, happy bday to the little guy |
| # | 12:18:32 | dbs | djfiander: SIPtest.pm is where $item_diacritic_title is defined, though |
| # | 12:18:47 | djfiander | ah. It's been a while since I looked at that stuff. oops |
| # | 12:18:52 | dbs | berick: thanks! and a merry christmas to everyone else on or off-channel |
| # | 12:19:29 | dbs | djfiander: yeah. but pines has it working somehow, so it's gotta work |
| # | 12:19:43 | djfiander | exactly. |
| # | 12:20:01 | dbs | okay, really going. Before I become a victim of spousal violence. |
| # | 12:20:06 | dbs has quit IRC |
| # | 12:20:06 | djfiander | because the same thing happened there: connected to an appliance. appliance started reporting problems. |
| # | 12:44:48 | djfiander has left #Evergreen |
| # | 13:19:30 | Dmagick has joined #Evergreen |
| # | 13:54:31 | bshum has quit IRC |
| # | 14:01:04 | phase_bb has joined #evergreen |
| # | 14:32:07 | gdunbar has quit IRC |
| # | 15:00:05 | alxp has quit IRC |
| # | 15:09:37 | jeff | By my understanding, and by a quick scan of the overdue fine generation code in 1.4, there's no current way to put a per-patron ceiling on how much a patron will be fined, there's just the per-circ maximum. |
| # | 15:12:27 | jeff | Is this correct, and has any work/thought gone into a ceiling on how much a patron will be fined, which could stop fines early if they have a large number of items out? |
| # | 15:13:37 | jeff | (If you have a 40 item limit and want $40 to be the max a patron could be charged for overdues, you could set a $1 per item/circ limit on the overdue rule, but then a patron with only one item long overdue would only owe $1.) |
| # | 15:14:11 | jeff | This seems to be the kind of thing that has come up before. I'll do some searching, but would appreciate any feedback in the meantime. |
| # | 15:32:29 | dchristens has joined #evergreen |
| # | 15:43:28 | dchristens | It's been almost a year since I did a batch load of bibs/items... followed my old notes, bibs loaded fine, copies look ok in the db (used the staging_table approach). Searching in opac finds the bibs, and shows # copies (in "title results") for local library system, but no copy information in "title details"... |
| # | 15:43:54 | dchristens | In staff client, I can bring up items by barcode. Any thoughts on where to start poking? |
| # | 15:52:01 | dchristens | Heh. I guess folks are busy :-) Happy holidays, all! |
| # | 16:13:46 | dchristens has left #evergreen |
| # | 16:17:29 | finnx has quit IRC |
| # | 16:18:22 | finnx has joined #evergreen |
| # | 16:20:37 | sfortin has quit IRC |
| # | 16:30:46 | leed has quit IRC |
| # | 16:48:13 | _bott_ has left #evergreen |
| # | 17:49:24 | mck9 has quit IRC |
| # | 18:12:02 | Dyrcona has quit IRC |
| # | 22:35:43 | alxp has joined #evergreen |
April 10-13