00:01:56 benadida (~benadida@66-108-96-200.nyc.rr.com) has joined #openacs 00:02:00 hello ben 00:02:14 hey there 00:02:32 I'm just trying to get some links and names correct for the docs... 00:02:41 ok 00:03:14 For example, where are we going to keeps the docs? The current site has 3.x docs in /doc/openacs but openacs.museatech.net has them under /oacs-docs 00:03:27 Maybe we should just have a /developer section and have everything there? 00:03:37 well, docs are not only for develoeprs 00:03:44 true. Just thought of that. 00:04:00 we need to have something very simple for docs 00:04:11 /documentation ? 00:04:13 You should point people to openacs.org/doc 00:04:24 and from there we will have a dispatch of all docs (including 3.x) 00:04:34 probably openacs.org/doc/3 and openacs.org/doc/4 00:04:37 Okay, sounds good. 00:04:52 we will make sure /docs and /documentation also work 00:05:06 but let's make /doc the canonical one 00:05:25 One item in the OACS FAQ is training. Before it listed the bootcamps, etc. Should I just point that to the OpenACS companies on http://openacs.museatech.net/community/ 00:06:09 an interesting question. Yes, pointing to companies is the right thing 00:06:48 adding a paragraph about how, in the exuberance of the dotcom boom, training for this was free, but now you'll probably have to get training at a certain cost. 00:07:06 unless of course there are people providing free training out there, but I don't know of any 00:07:21 me either 00:07:49 There are lots of links to the (defunct?) ASJ. Are we going to mirror that on openacs.org? 00:08:08 we need to get explicit permission from the authors, technically. 00:08:13 cause that's redistribution. 00:08:25 I would like to mirror it, yes, that would be nice. But we need permission. 00:08:40 Indeed. 00:08:41 can you contact some of the authors? 00:09:02 One guy just wrote me about helping with documentation. I could delegate that to him. 00:09:29 that works. Make sure this is done politely, we are asking for permission, not demanding it :) 00:09:54 Okay. Got it. 00:10:02 It looks like arsdigita.com/asj is still up 00:10:23 yeah, and if we're really stuck, we *can* point to the web archive site (that site is unbelievably cool, btw) 00:10:48 archive.org? Yes, I found out about that the other day. It's super. 00:10:55 yes, that's the site 00:11:28 Another frequent thing is the ArsDigita Server Architecture. 00:12:32 what is that about again? Is this Philip's discussion of keepalive and all? 00:12:42 Yeah 00:13:46 and what's the conflict? 00:14:22 Well, it used to be at arsdigita.com/asj 00:15:11 oh, then do the same thing as for ASJ: ask the author (in this case Philip) if we can mirror it 00:16:14 k 00:16:44 in fact, where Philip is concerned, I'll email him now and ask. 00:18:09 I'll keep you posted on his response 00:19:33 okay 00:19:54 I'll be chugging along, so if you could stick around for a bit, that'd be great. 00:20:42 I'll stick around, but this window doesn't seem to grab focus when I go away, so you might not get immediate responses 00:21:05 benadida: That's okay. I'll just bug you through AIM if you take too long to respond :-) 00:21:39 benadida: Thanks for the help 00:26:16 re 00:31:08 ok, so now I have a pair of wimpys (just like before) but now they have a presentation apiece... I had to apply my patch to get them to restrict presentations to their respective creating package instances 00:31:25 jim: nice. 00:31:44 now, trying to delete the package 00:33:55 I've found that there are a lot of packages that are not scoped correctly. 00:34:02 benadida: Should the links to amazon.com on the oacs docs still reference "photonetA"? (I guess Amazon pays them a commission for that?) 00:34:08 even bboard has issues (where alerts are concerned) 00:34:28 no, that's probably not cool -- make the links non-commission for now 00:34:37 k 00:34:59 jim: i'm back too 00:35:20 got errors. summary so far: wp-slim's delete script works when: 00:35:36 - no instances defined or mounted 00:36:00 - mounted instances exist with no presentations 00:36:58 just now discovered... drop script gets a small handful of errors when there exist mounted instances that have otherwise-empty presentations 00:37:08 re davb 00:37:45 end of summary 00:38:09 dropping openacs instance, recreating... 00:38:47 benadida: And do we want to have a "/devel" or "/developer" section on the website or do we just want to continue with "/contribute"? 00:39:30 I think /developer is a good idea 00:44:56 okay. 00:46:10 donb (~donb@dsl-dhogaza.pacifier.net) has joined #openacs 00:46:36 Roberto - you around? Got your e-mail and here I am (howdy Ben, Dave, etc) 00:47:26 ok, new test oacs instance installed, wp-slim installed. 00:47:28 hello Don 00:47:37 hi don 00:47:55 hello everyone else (wow, look at all those people, just got my Fire client doing the right thing) 00:48:57 I just e-mailed Roberto a quick response to his question, Ben, suggested he just ignore that particular issue for now and in the future we can split off a section saying something like "ArsDigita used to recommend ..." for such historical stuff. 00:49:29 Of course, we can redo it or do our own if there's time/energy but I sure don't want to hold off beta for doc issues of this sort. 00:52:07 (note about stability... openacs will be viewed as more stable if, generally, deleting things always works... overall package drop scripts, deleting package instances, deleting objects created by instances. 00:52:10 ) 00:53:19 this will also help in debugging, if you can "tinkertoy" it: put things together, take them apart, without destroying and recreating the tinkertoy set :) 00:53:27 Yes, Jim, you're right, and this is one reason I lobbied aD hard trying to convince them to use proper referential integrity tools to maintain, well, referential integrity ... i.e. "on upate/on delete" referential integrity actions. 00:54:14 SQL92 has perfectly good built-in tools to help with this ... and both Oracle and PG implement them (PG implements more of them, actually). 00:54:39 as it is now, if I get into a weird situation, I think I have to delete all the oacs and start over from scratch 00:55:06 hey don 00:55:18 (this is not true in actual fact... I could piece together the things needed to remove if only I knew the data model :) 00:55:43 donb: I've been asking some questions for Ben. 00:56:27 but since I don't know the data model well enough, right now it's easier for me to blow the whole thing away and starting over, especially if I'm trying to debug a little piece 00:56:29 brb (be right back for the IRC newbies :-)) 00:56:53 Jim - I find myself swearing and dropping/recreating everything from scratch far too often. aD uses RI actions (on delete etc) sometimes, and not other times, it makes things a mess. 00:57:31 I know the datamodel well and it doesn't really help that much, because you have to track down all the references threading things togther and have to delete things in the right order. And a lot of the drop scripts don't work if there's data. 00:57:42 was it to "get it out the door" or "get the site up for the client" kinds of reasons? 00:58:17 Philosophical, on delete and on update RI actions were considered "too dangerous" - I think they must've read the MySQL docs :) 00:58:21 donb: drop scripts not working if there's data, that's -got- to change 00:58:29 "get the site up for the client" implies there is a client site running ACS 4.x 00:58:39 Yes it has "got" to change ... 00:59:10 we're working on wp-slim right now... 00:59:23 Trying to get the drop scripts to work or something else? 00:59:57 davb and I created a situation where now presentations are tied to their respective package instances 01:00:32 it seems to work pretty well, but I don't know if I've covered every place where wimpy creates a list of presentations 01:00:59 Oh, that's good ... are you doing this for Oracle and PG or just one? 01:01:06 both 01:01:53 oh, question: should a "presentation title" be allowed to contain html tags? why or why not? 01:01:53 Cool ... if you think you're done and test it well we can slip it into beta, but you're going to have to hurry and convince me you've tested adequately. I'd kinda like to start rolling the beta tomorrow (my birthday) ... 01:02:17 Don, your bday is tomorrow? 01:02:23 Yep, tomorrow ... 01:02:24 my bday was 4 days ago :) 01:02:25 happy bday 01:02:31 and my sister's is tomorrow :) 01:02:32 and happy bday 01:02:40 and happy bday yet again... 01:02:46 Oh, really, Ben? Wow, that's pretty cool. 01:02:51 all the february bdays 01:02:59 aquarious? 01:03:37 I probably have as many bdays as you and your sister put together, damned near ... pisces, Jim ... and I haven't thought about html and presentation titles, probably no big deal *not* to have html allowed ... 01:03:59 yeah, but youth is all in your head, Don :) 01:04:13 see tho, I don't know yet how to test adaquately 01:04:48 donb: ok, I'll patch to that effect... 01:05:42 (the thing about the presentation title is it is itself an html link, not sure how links inside that would affect it) 01:06:01 jim: sounds like you should keep it plain text for sure, then. 01:06:12 jim: not sure what a link inside the title would do anyways 01:07:15 as far as testing so far... 01:07:41 docwolf (~docwolf@adsl-34-207-203.bct.bellsouth.net) has joined #openacs 01:07:50 let's see, I could post that summary... hasn't changed since then 01:08:07 jim has quit (Remote closed the connection) 01:08:56 jim (~jim@12-233-187-5.client.attbi.com) has joined #openacs 01:09:13 oops, just lost my scrollback 01:10:32 Jim - check out "ad_looks_like_html_p" in acs-tcl/tcl/text-html-procs.tcl. It takes a stab at telling you if a string contains HTML or not, and you can flag the title with an error if so. There's a lot of good stuff in that library and I've been folding stuff into various packages as time permits. 01:11:21 found the summary... 01:11:24 00:35:20 got errors. summary so far: wp-slim's delete script works when: 01:11:24 00:35:36 - no instances defined or mounted 01:11:24 00:36:00 - mounted instances exist with no presentations 01:11:24 00:36:58 just now discovered... drop script gets a small handful of errors when there exist mounted instances that have otherwise-empty presentations 01:11:25 00:37:08 re davb 01:11:27 00:37:45 end of summary 01:12:38 I think we can wait 'til post-beta to clean drop scripts. They're a PITA to get right but we really need to get this beta out there. 01:13:03 Of course if you get me something in the next day or so I'll take it, never fear! :) 01:13:30 also, when the drop script has been run if there are mounted instances, going to the site map, you'll find them still mounted, going to their url will produce an error, unmounting and deleting the instance works tho. 01:15:10 what I was wondering tho, was this: (1) is it possible in sql or plsql to determine if a package instance is mounted? (2) if so, can site map be told by the drop script to unmount then delete the unmounted instance (without deleting the mount point)? 01:15:44 that would be slick, and that would be a place you wouln't get an error anymore 01:17:03 You can tell if a package has mounted instances (Ben should be able to do this in his sleep by now because of dotLRN, I'd have to go poke around the datamodel, actually I think some of the site-node utility procs OF did for dotLRN will do most or all of the check, it will go in the main tree at some point) 01:17:09 (while dropping a package should clean up its instance and package stub in the site map, the drop script has no business altering the site map structure) 01:17:23 drop should not modify the site map structure, absolutely. 01:18:29 I was thinking, if you have in the site map a row that looked like this: 01:19:14 wp/ Wimpy 01:19:38 and then you delete the Wimpy, 01:19:54 then the site map should look like this: 01:20:02 wp/ (none) 01:20:43 Yep ... or the user should be forced to unmount the packages before dropping and get an error from the script otherwise (could provide something in the APM API to check that drop scripts could call). 01:21:00 "could" not "should", I mean, haven't really thought about this issue . 01:21:13 hey donb 01:21:17 donb: if you think of it like a mac application... 01:21:27 whoa. and benadida 01:21:29 hubba 01:21:49 you should allow that to happen, and perhaps let them know, asking for an "are you sure" 01:23:25 "are you sure" is hard to do in PSQL scripts ... 01:23:50 hmm 01:24:00 Now an APM wrapper to run the drop scripts for you could check for mounted packages, do the unmounting first (rather than burdening the packages with this chore) and then call the drop script just as it now calls the create script. 01:24:07 and changing that would involve radical changes 01:24:34 or altering the structure of the oacs to incorporate that ability 01:24:58 Driving it from the APM is consistent with how packages are installed ... that's a good thing. 01:26:08 maybe the apm could get some extra buttons for developer purposes... "drop datamodel" and "reload datamodel" 01:26:55 * jim got the 80g drive from aduni today :) 01:28:15 what's all the special fun in here for? 01:28:26 are these donb and benadida characters new bots? 01:29:48 Hey, I've been here before! Jim - a good developer button: "drop the datamodel worldwide!" Ka-boom! What fun! 01:31:04 and next we'll need a "install ossama GPS"? :) 01:31:10 I gotta run - rbm, if you need more doc feedback e-mail me again and I'll answer later today ... 01:31:12 but donb, you have a girlfriend. why are you here? 01:31:27 the only people here either haves wives or are single. either way, we have no life 01:32:00 sorry married guys. i don't mean to speak for you 01:32:09 i know thats your wives' jobs 01:32:38 We know Talli has no life ... :) We're going to have fun at the Lucky Lab on Wednesday and you aren't! 01:32:47 whoa. that's low 01:33:00 And now I'm really leaving, so there! 'til later ... 01:33:02 donb has quit ("changing universes") 01:42:37 ok, running the next test... 01:48:03 this is a new user, created after the wimpy instance was created and mounted, creates one presentation with no slides. 01:48:53 now, we'll delete the whole package, then unmount and delete the stub (full log of "delete the whole package" will be produced) 01:52:00 208 psql:wp-slim-drop.sql:143: ERROR: acs_obj_context_idx_anc_id_fk referential integrity violation - key in acs_objects still referenced from acs_object_context_index 01:52:14 in the drop script's inline5() 01:53:02 295 psql:wp-slim-drop.sql:208: ERROR: cr_items_rev_type_fk referential integrity violation - key in acs_object_types still referenced from cr_items 01:53:09 in inline_8() 01:53:37 334 psql:wp-slim-drop.sql:245: ERROR: cr_items_rev_type_fk referential integrity violation - key in acs_object_types still referenced from cr_items 01:53:43 in inline_20() 01:53:49 jim, is the problem with wimpy point a big problem or just one with wimpy point? 01:54:13 well.. 01:54:44 jim: i think we need to tackle them one at a time. if the first transaction rolls back, it will screw up the later ones because something wasn't deleted... 01:54:46 see, these problems are correctable in wimpy point itself... but instances of them exist in lots of the drop scripts 01:55:02 ok 01:55:27 208 psql:wp-slim-drop.sql:143: ERROR: acs_obj_context_idx_anc_id_fk referential integrity violation - key in acs_objects still referenced from acs_object_context_index 01:55:32 that's the first one 01:55:37 for inline_5 I think the answer it to break it into two functions. first go and turn all the context_ids to null. 01:55:43 then end 01:55:50 then go back and delete all that stuff. 01:56:45 I am stuck on the fact that there is no way to ask a package what it's URL is. 01:57:16 davb: Are you sure? I think there is. hang on 01:57:21 you should be able to get a list of package instances, and ask the site map which ones are mounted, and if so where 01:57:23 cool 01:57:40 a package is usually not itself mounted 01:57:43 ok I am stuck on the fact that I can't find the proc that does it :) 01:58:07 one question tho... 01:58:11 right. a package instance. I will have several ETPs I want to generate a navigation for all of them on the home page. 01:58:56 if we're now using context_id in the way it was designed for, why have code that treats it specially? 01:59:04 davb: set package_root_dir [acs_package_root_dir srl-sds] 01:59:07 pg bug 01:59:13 where "srl-sds" is your package name 01:59:22 cool. thanks 01:59:32 I wonder if its fixed in 7.2 01:59:46 either that or I am wrong... 02:00:47 let me try it in pgsql... 02:01:15 I mean, couldn't we now just call acs_object__delete? 02:01:43 well we do. cr_item__delete calls acs_object__delete :) 02:02:35 I had to write a custom acs_object__delete for static_pages drop script to work... 02:04:59 ah. same error in psql 02:05:11 update acs_objects set context_id=NULL where context_id=2597; 02:05:17 ERROR: triggered data change violation on relation "acs_object_context_index" 02:05:28 that is what is causing the error. I am wrong :) 02:05:49 maybe the trigger is broken? 02:06:06 I wonder if any other package tries to unset a context_id like that. 02:08:39 its happening somewhere in here I think: acs_objects_context_id_up_tr 02:12:54 * jim starts reading at line 416 02:16:46 I don't get where it's getting old and new 02:17:09 which probably means I forgot my understanding of triggers 02:17:49 ah. magic. old is before the update I think, and new after. 02:18:39 the problem is PG won't tell you where the error occuring in a nested function call. 02:20:38 insert loggy thingys? 02:21:21 yeah, that is all we can do. drop the trigger, then the function and recreate the function then the trigger :) 02:21:45 ok, signing out guys 02:21:47 talk to you later 02:21:59 benadida has left #openacs 02:22:06 nite ben 02:22:11 2l8 02:22:13 jim. this may be the wrong way tot do it. 02:22:35 no other packages seem to set context_id = null... 02:22:36 you mean determine the problem? 02:22:48 that's exactly what I was thinking 02:22:49 no to delet ethat stuff.. still looking 02:23:12 I am going to check out some other drop scripts, maybe bboard... 02:23:17 before, you had nonstandard use of context_id,,, yes? 02:23:33 bboard drop script doesn't work either 02:23:40 ah. that helps. 02:24:45 No, it just left out the context_id or parent_id of the presentation so it didn't belong to any pacakge. 02:26:22 so if it did that, how did the script fail before? 02:26:45 it is the same error we were getting beore last time we tried this :) 02:27:23 ok, what if we commented out the code that does special things with context_id on delete? 02:27:51 that is what I was thinking. we might get another error, but we would see why they thought they had to set context_id=null. 02:28:15 i.e., let the context id be set to the package instance when the acs_object__delete() happens 02:28:34 hey it worked. 02:29:00 cool :) what file/line? 02:29:06 no, just leave it alone actually. there is a delete trigger also 02:29:17 I just did content_item__delete(item_id) 02:29:22 for a presentation... 02:29:49 except I got this error: ...ERROR: Relation 'cr_wp_slides_preamble' does not exist 02:29:56 not sure why. 02:30:04 so in that case the context id was set 02:30:14 yes. I didn't unset it. 02:30:47 I need to get the patch for acs_object__delete in. 02:31:03 that is the custom content_type that doesn't has its own table error. 02:31:19 it should just do a table_exists(tablename) before trying to delete it. 02:31:30 I though I submitted that...let me look... 02:31:48 question about that... could the probmel also be solved in the cr package? 02:32:22 very likely. 02:32:41 (oop ism: a superclass shouldn't have to be special to handle its subclass) 02:32:53 see cr_item__delete calls acs_object__delte etc... so if you function calls another function you never know exactly what caused the error. 02:33:17 well I think we might just not bother trying to update context_id at all. 02:33:34 I think someone mistakenly believed you needed it. (and it might not hurt in oracle) 02:34:22 that was a preamble, not a presentation. but still if the acs_objects works. 02:34:51 I thought my acs_object__delete hack was just for etp, but it appears other packages broke the cr rules too. 02:35:51 if table_exists(obj_type.table_name) then 02:41:03 ok, what are those rules? 02:41:17 let me reload my data model and see if my patch fixes that... 02:41:23 otherwise we will be stuck. 02:41:40 bakataro (~rolf@above.badgertronics.com) has joined #openacs 02:41:54 yo bakataro 02:41:58 yo. 02:42:14 i heard you may have acquired a nice piece of hardware today 02:42:38 yes it is pretty nice. 02:42:49 now if i could figure out the fscking menu in OSX 02:42:50 er 02:42:54 photshop 02:42:56 bummer 02:42:57 +o 02:43:03 you get any software for it? 02:43:09 * davb knows photoshop ok... 02:43:25 is Office vX have a crappy rent license like all other MS software now? 02:43:34 davb: do you know how to get osx to display the photoshop menu properly 02:43:37 of course, I am not running a photoshop friendly os right now... 02:43:40 under carbon emulation. 02:43:51 ah. that is beyond my meager skills sorry :( 02:44:05 office vX is a normoal license so it seems. 02:44:10 hm. 02:44:32 rent license?? 02:44:44 wow, the datamodel loads really nice on this new machine 02:47:16 openacs4=# select content_item__delete(2595); 02:47:17 NOTICE: Deleting symlinks... 02:47:17 NOTICE: Unscheduling item... 02:47:17 NOTICE: Deleting associated revisions... 02:47:17 NOTICE: Deleting associated item templates... 02:47:17 NOTICE: Deleting item relationships... 02:47:19 NOTICE: Deleting child relationships... 02:47:21 NOTICE: Deleting parent relationships... 02:47:23 NOTICE: Deleting associated permissions... 02:47:25 NOTICE: Deleting keyword associations... 02:47:30 what on earth. 02:47:30 NOTICE: Deleting associated comments... 02:47:32 NOTICE: identifier "journal_entry__delete_for_object" will be truncated to "journal_entry__delete_for_objec" 02:47:35 NOTICE: Deleting content item... 02:47:36 content_item__delete 02:47:38 ---------------------- 02:47:40 0 02:47:42 (1 row) 02:47:44 tada 02:47:47 I created a presentation, and just did content_item__delete(item_id of the preamble) 02:47:49 sorry. 02:48:10 jim: let me make the acs-kernel-patch you will need to make it work. 02:48:39 davb: ok 02:49:34 ok I just deleted everything with the presentation by doing cr_item__delete(item_id of presentation); 02:49:53 that would simplfy the drop script a little 02:51:12 * jim waits for script... wants to see if it can go into cr package 02:51:18 err, waits for patch 02:51:54 the patch fixed acs_object__delete for when someone cheats and doesn't create an attribute table for a custom object_type. 02:54:00 ok, so: they make an acs_object, and an object type, and let the system create the table/ 02:54:01 ? 02:54:51 right, but the package doesn't use the table to store attributes, so it is not created. all I do is check if a table exists before trying to delete it. 02:56:11 I can delete a presentation and all assocaited content with one call to cr_item__delete with the item_id of the presentation. at least is psql. 02:56:28 the doc links aren't all working... 02:59:51 which one(s)? 03:01:45 benadida (~benadida@66-108-96-200.nyc.rr.com) has joined #openacs 03:01:53 I'm at /doc/object-system-design.html 03:02:01 benadida has left #openacs 03:02:13 ah. I thought they worked before... 03:02:35 trying to click on those three sql files, one of them is acs-metadata-create.sql 03:02:46 none of those work 03:03:21 ah. 03:03:31 where do they link to? 03:03:49 I commented all the set content_id = nulls and got to here: psql:wp-slim-drop.sql:416: ERROR: RemoveFunction: function 'wp_presentation__new(timestamp, int4, varchar, varchar, varchar, varchar, int4, bool, bool, varchar, varchar)' does not exist 03:03:57 heh, that was my other patch... oops 03:04:56 right, that should work 03:05:00 I think we were looking in the wrong place all the time! 03:05:21 instead of figuring out why the set=null was failing, it shouldn't have been there in the first place :) 03:05:42 we were looking in so many different places, I forgot which ones were rigth or wrong :) 03:05:47 now that my drop script half-ran I will have to reload the data model 03:05:49 yes, I suspected that 03:06:02 I think your original patch removed the reason for it 03:06:18 ah. all that extra work... 03:06:29 ok. let me try one more time. 03:06:46 I also comments out the delete from cr_item_audit; 03:07:11 they were too busy to make a where clause, so why not delete all the rows and leave a comment -- do not leave this 03:07:13 dunno even what that is... internal table? 03:07:37 audit trail for every time a cr_item is modified. every cr_item that way they had it :) 03:07:48 not sure how to fix it. 03:08:20 I think cr_item__delete should handle it for you... 03:08:26 it seems to me the cr should take care of that... deleted item should take the audit trail with it 03:08:28 of course, on delete cascade would to :) 03:08:40 well I commented it out and it still worked. 03:09:03 hmm, /me thinks that's debatable, given possible desire to archive audit trails... 03:09:24 the item_id refers back to the item that does not exist anymore though. 03:09:43 I doubt there is a lot of package expunging like this on a live site. 03:10:25 unless you just don't have a foreign key on that column in the audit table... 03:10:39 but people experimenting and "furniture rearranging" have to have a solid ability to delete/create/delete/createsomewhereelse etc 03:11:14 right. 03:11:15 well, depends on what this audit trail is used for 03:11:23 right. 03:11:57 I think new cr items can decide to archive the audit trail upon delete if that's what's needed 03:12:23 talilee has left #openacs 03:12:28 right. 03:12:32 or left 03:12:33 :) 03:12:49 this new machine is an order of magnitude faster on loading the datamodel at least. 03:12:57 the extra ram must really help. 03:14:24 well running X, mozilla, xchat etc... doesn't leave much left for aolserver/postgresql... 03:14:57 yep. I am using around 320mb, the exact amount fo ram I used to have. 03:16:09 I' 03:16:17 I'm getting a problem... 03:16:27 ok. 03:16:37 I might be in some weird position, gonna load dm again 03:18:28 successfully droped one instance with one presentation. 03:19:40 BTW, the mount point of the instance is gone. I didn;t do anything special to it. 03:22:59 the mount point? 03:23:02 interesting 03:23:15 ok, I wanna get to where you are 03:24:19 all right. 03:24:32 if I blow away my wp and get from cvs... 03:24:45 I'll have your patch that sets the context id 03:26:43 right. 03:26:50 bakataro has quit ("bakataro has no reason") 03:27:01 and won't have my html patch or my wp query patch 03:27:17 (just checked for the html one) 03:27:30 right. you also need the acs_object_delete patch I sent. 03:27:49 do you want a diff of the drop script? 03:27:57 gimme a sec 03:28:35 since the other alteration to the html isnt' too hard, I wanna do that now 03:31:28 is a unified diff ok? 03:31:41 for openacs.org or me? 03:31:42 or they absolutely need context diffs? 03:31:50 openacs.org 03:32:14 I think they prefer context diffs with one bug each. easier to apply, but they aren;t super picky. 03:32:43 multiple patches per bug ok? 03:33:20 patches per bug? yes. bugs per patch is ok too. you can assign a patch to more than one bug in the sdm. 03:33:23 or should each succeeding patch on a single bug do everything plus newly discoveres stuff? 03:33:38 ah. I am not sure. 03:33:40 rbm? 03:34:03 ok, gonna have 2 patches then, they can always be combined into one file ;() 03:35:14 created the patch then restored myself to just what's in cvs 03:35:30 I need my context id patch... yes? 03:35:52 for the queries? 03:36:11 right 03:36:48 it probably doesn't affect the drop script, but I have it. 03:39:06 argh, ridley scott is reediting bladerunner and I just got the dvd. 03:40:10 hmm... you're gonna submit the patch once we have this together? 03:40:36 for which? the drop script? sure 03:40:52 (I just submitted patch 163, which together with 123 sew up the html thing) 03:44:27 hmm... I haven't installed wp yet... should I apply these patches first? 03:44:39 also... 03:44:43 if I' 03:45:03 if I'm applying a patch to acs-object, should I reload data model? 03:45:26 yes. 03:45:48 to the first q, ok, looking for the patch 03:47:11 yes apply the patches before installing wp. 03:56:10 but I don't need to install data model? 03:56:27 that too. 03:56:28 i.e., of whole acs 03:56:46 yes if you apply the acs-object-delete patch you need to reload the datamodel 03:57:07 right, because you're altering a stored proc 03:57:13 got it 03:57:23 ok, before I do that... 03:58:17 what about the specific changes to wp? 03:58:35 you said something about handing me a patch for that 03:58:47 ah... 04:00:11 ok sent 04:00:51 to me? or the sdm? 04:03:58 not in sdm, so checking email 04:05:18 email first. 04:06:03 if it works for you, I'll send it in. 04:07:25 should I actually delete out all those lines for the real patch? 04:09:54 which ones? you're commenting things out instead? 04:10:24 I remember saying we should comment to start out with 04:10:48 lemme reinstall this thing 04:10:52 right. 04:13:58 ok, reinstalling data model 04:14:14 talilee (~talli@lti-4.dialup.access.net) has joined #openacs 04:19:53 ok, ready to run drop script 04:20:58 * davb holds breath 04:22:35 no errors 04:22:42 cool! 04:22:46 yep :) 04:23:18 after I did the drop. I restored the files, reinstalled back over the same direcotry and did it all again and it worked. 04:23:36 probably should check out the oracle side too :) 04:23:38 so, now we have presentations existing, no slides, deleted fine. 04:24:10 lemme finish pg first :) there are more conditions to test :) 04:24:22 I wonder if all the set context_id = null stuff is for oracle only. 04:24:39 naww, oracle should be fine 04:24:45 right. 04:25:08 oh yeah, I also installed two instances and put prensentations in both and dropped sucessfully. 04:26:59 check other packages too, like, which other package besides wp-slim had problems/ 04:27:00 ? 04:27:31 and check some that didn't have problems 04:27:44 or that we didn't know or deal with so far 04:29:02 ah good idea. 04:29:18 well the first patch fixed the etp/wp-slim conflicts 04:31:29 cool 04:32:17 ok. I will be back tomorrow. are you going to submit the patch to fix the queries to restrict the list of presentations? 04:33:33 * jim checks to see if he did that 04:34:43 patch 161 04:35:04 ok. cool. I think the only other patch is the one for the actual drop script then right? 04:35:25 (and 160, but I wanna delete that one cause it's a bad version of 161) 04:35:45 right, the wp-specific one 04:36:42 ok. good. that was easier than I thought it would be. 04:36:52 my patch 161 does oracle too, btw... was an easy copy-it-in-there patch 04:37:16 once you look deep enough, most things are pretty easy 04:37:29 oh yeah. i need to see if acs_object.delete for oracle needs a table_exists() or whatever oracle uses to check table existance. 04:38:13 oh, I dunno how to check for table existance in oracle 04:38:21 I will find out tomorrow :) 04:38:32 in pg, the table names are in a table :) 04:38:32 good night 04:38:39 nite. damn good work. 04:38:50 np, it was luck :) 04:38:54 davb has quit ("Client Exiting") 06:55:26 dlk (dlk@walter.ita.chalmers.se) has joined #openacs 06:55:26 talilee has quit (Read error: 104 (Connection reset by peer)) 07:31:50 I just got an interesting error trying to delete a package instance... 07:32:12 [24/Feb/2002:23:31:19][29511.49158][-conn7-] Warning: APM: PersistentLoginDefaultP does not exist 07:32:12 nsthread(30340) error: ns_malloc: could not allocate 567363840 bytes 07:32:12 Aborted 07:32:12 dual:/usr/local/aolserver> 07:32:46 why would it think I even had that much?? 07:33:23 maybe it read my mind and it knows I have 512 installed and another 256 mb ready to install... 08:08:41 ce (~ce@stalker.ez.no) has joined #openacs 09:59:05 kidding there, ofcourse... 10:34:23 talilee (~talli@lti-4.dialup.access.net) has joined #openacs 11:59:27 PascalS (~pascal@facb-167-233.pc.fgg.eur.nl) has joined #openacs 12:09:25 ce has quit ("My damn controlling terminal disappeared!") 12:23:04 til (~tils@port-212-202-128-195.reverse.qsc.de) has joined #openacs 12:57:57 dlk is now known as dead-man-walking 14:16:27 davb (~dave@rrcs-nys-24-97-22-203.biz.rr.com) has joined #openacs 14:18:50 hello 14:21:20 morning dave 14:21:38 http://www.sinc.sunysb.edu/Stu/nblyumbe/deskmod 14:21:47 heh..Chump is not here 14:24:32 argh. 14:24:57 be back later 14:25:00 Bronx time 14:25:14 chump (~chump@alb-24-58-160-41.nycap.rr.com) has joined #openacs 14:25:17 have fun! 15:06:10 k2pts (~nkd@adsl-168-174.cytanet.com.cy) has joined #openacs 15:06:15 hey guys 15:06:20 davb: are you around? 15:06:28 paje: seen davb? 15:06:28 davb was last seen on #openacs 41 minutes and 11 seconds ago, saying: have fun! [Mon Feb 25 07:27:12 2002] 15:06:47 til: are you around? 15:07:00 yes 15:07:14 i've got a question for you about one of your patches, moment 15:08:38 sure 15:08:46 http://openacs.org/sdm/one-patch.tcl?patch_id=116 15:08:46 A: http://openacs.org/sdm/one-patch.tcl?patch_id=116 from k2pts 15:09:07 the context_bars, why is that needed? why don't we use ad_admin_context_bar instead? 15:10:06 btw, do you know where I can find vinod's patches for ns_uuencode 15:11:19 i never heard of ad_admin_context_bar before, maybe that's why i didn't use it ;) 15:12:11 ok I'll check it out later this week and apply it 15:12:24 if ad_admin_context_bar cannot replace that functionality 15:12:49 do you know if I can find vinod's patches somewhere in a tarball? 15:13:57 sorry, i never used them, so i don't know where to find them. 15:14:03 ok til, thx 15:14:14 have to head out, l8r 15:14:20 k2pts has left #openacs 15:25:55 * davb submits the wp-slim drop script patch to the sdm 16:05:32 PascalS has quit ("Client Exiting") 16:09:37 dead-man-walking has quit (Read error: 110 (Connection timed out)) 16:24:08 hey guys 16:26:00 hi talilee 16:40:08 http://www.easypooltutor.com/ 16:40:08 B: http://www.easypooltutor.com/ from davb 16:40:14 B:|Online pool lessons 16:40:14 titled item B 16:44:33 talilee: any news on the openacs 4 migration? 16:44:44 you mean the new site? 16:44:58 yes 16:45:02 i have an emacs window open with the task list. i need to finish it, but it will be done today 16:45:05 openacs.org->4 16:47:03 ok. wow, I didn't know you used emacs! 16:47:33 other than the bboard migration and setting up the download section, we need to customize the news module for a few sections. rather, we need to set something up so that people can submit new sites, news, new companies, events, etc. for an admin to accept 16:47:49 i use emacs on winxp for most of my HTML and text editing 16:48:02 cool. 16:48:19 i would use vim if i could, though. 16:48:56 no vim for win? 16:49:05 uhm, well, it's annoying 16:49:22 ah :) 16:49:40 there's absolutely no GUI for it, and windows is kinda tough to use without a GUI 16:49:57 especially since the folders/directories suck to have to navigate 16:51:34 I can't get used to vim. 16:52:11 it takes a while, but i really prefer the different modes 16:53:24 there must be a vim mode for emacs :) 16:53:39 there is! 16:53:43 it's called viper 16:53:46 it's really cool 16:53:57 moo 16:54:10 hey rbm 16:54:10 hi rbm 16:54:47 talilee: There's vim-gtk for windows. It's a graphical vim with menus and all that. 16:55:32 * rbm installs cygwin in all his windows boxen (only one for his wife :-P) 16:55:49 talilee: I have some news for your wrt to the new openacs.org 16:56:19 I discussed with Ben and Don some things while I was editng docs yesterday and we reached some conclusions wrt to URLs and site areas. 16:56:28 yeah, i saw that 16:56:33 (I actually discussed it here in the channel with everybody that was here) 16:56:42 i came in right afterwards 16:56:43 talilee: Oh, great then. 16:57:11 it doesn't really matter what it is on openacs.museatech because it should move to the OACS box asap 16:57:37 yeah, you should move that stuff over and finish developing on the OACS box. 16:57:47 talilee: But whatever is on openacs.museatech will move to openacs.org. 16:57:58 true enough 16:58:50 * rbm just finished his homework!!!!! 16:59:04 * rbm sacrifices a lizzard to the college gods 16:59:48 * rbm ponders buying a #!/usr/bin/girl tee for his wife 17:02:18 til has quit (Read error: 113 (No route to host)) 17:09:24 rbm, i have some baby-t openacs4 tshirts 17:09:33 for the girls 17:09:54 talilee: No way! 17:10:03 talilee: I want one :-) 17:10:04 i do. only ten of them 17:10:18 i don't think it would look very good on you rbm. i think it would be for your wife. 17:10:33 talilee: :-P Obviously 17:10:33 i know you like to show off you're ample bosom, but... 17:10:46 Me? Ample bosom? Ha! 17:11:59 * davb is thinking about designing another logo for OpenACS 17:12:09 something with a bridge or something in it... 17:14:49 * rbm wishes Mozilla would be more like Mozilla 17:14:54 Ack 17:15:05 s/Mozilla$/Opera/ 17:16:15 really, I wished opera was more mozilla like sometimes :) 17:16:56 If the Mozilla team were trying to build a browser... But they're out to build everything but a browser. 17:17:33 ah that part. 17:17:53 but that may help me out. I want to build a rich browser based interface to openacs. 17:18:14 davb: ? What do you mean? 17:19:24 Basically a desktop app to access an OpenACS CMS such as ETP. 17:19:31 let me see if I can find a link to something similar. 17:20:20 ahh. That'd be cool 17:20:50 * talilee might sound like a broken record but... 17:20:52 webdav 17:21:00 right. 17:21:19 I am interested in Mozilla as the platform though. because of its freeness. 17:21:29 and we have to get in there soon or we'll fall behind a lot of the other packages 17:21:40 mozilla would be nice. but there's also emacs 17:21:46 but mozilla would provide a nice gui 17:22:07 don't know if I agree 17:22:34 talilee: do other toolkits really have real webdav support? 17:22:41 I was able to find few real libraries. 17:22:46 but there are also ubiquituous apps out there like word that might provide great access too 17:23:05 sure. word is on the list also. it has to be. 17:23:05 word, emacs, dreamweaver, golive, illustrator, indesign 17:24:03 talilee: getting graphics and stuff into openacs via webdav would be nice. but the writing tools are more important, I think, than the designer/programmer tools. 17:24:32 right. 17:24:50 but word, emacs, dreamweaver and go live are all great for writing as well 17:24:59 dreamweaver? 17:25:12 in addition, there's an app called soft metal XML that is webdav capable 17:25:28 dreamweaver can be used for writing content as well 17:25:36 sure, but how many clients was to create content in html or xml? 17:25:47 they don't have to 17:25:54 but all of those tools are WYSIWYG as well 17:28:59 ah. I am looking at neon. you have seen that? 17:29:05 yeah 17:29:31 it's a project to build a C library for webdav servers 17:29:38 it doesn't get a lot of traffic 17:30:22 looks like it has been active lately 17:30:51 a little. they just released a point version this week, i think 17:31:18 but the API is still changed 17:31:20 changing 17:31:25 yeah I saw that. 17:31:47 so i dunno how advanced it is. if it's a new project or just has been taking a long time. not that a changing API is always bad.. 17:32:44 right. I am not sure how much of AOLserver is actually recreated inside that though... 17:32:56 yeah, that's a good point 17:33:06 so neon is a whole server, or just the webdav part? 17:33:09 because I am not a C hacker at all 17:33:24 webdav, but it seems to have all the http stuff in it... 17:33:30 ah, i see. 17:33:49 wait, its a client library. wrong direction... 17:34:01 is it just clients? 17:34:06 i thought it was for the server too 17:34:28 could be :) 17:35:50 how hard would it be to retrofit mod_dav to work with aolserver? 17:36:19 talilee: I though tom@zbh had done some work in that area 17:36:44 in webdav? 17:36:51 yeah. webdave+aolserver 17:36:56 s/dave/dav/ 17:37:00 i don't think so. none that i've seen. he's done the jabber stuff, but i'm not sure about dav 17:37:11 Although "webdave(bauer)" would be cool :) 17:37:47 So with dav you can write any content to the server? Not just html? 17:37:58 yeah 17:38:17 webdav is pretty much secure FTP, via HTTP 17:38:35 its more. it includes versioning also. 17:38:45 right 17:38:53 but you can suck things out of a DB, too, right davb? 17:38:54 I can see it transferring files, but I don't see how it would work with something especific like etp 17:39:15 rbm: aolserver doesn't have to save the files in the filesystem. 17:39:40 it can hand them off to openacs... 17:40:07 gotta step out. bbl 17:44:02 ah mod_dav is being rolled into Apache 2.0 17:44:48 it has a nice friendly liscense 17:45:18 except for the name restriction... 17:45:57 davb: I'm looking at the source code for mod_dav and it doesn't look like it should be hard to port it to AOLserver 17:46:18 We'd mostly have to change what it sends and receives from Apache. That's it. 17:46:24 (apparently) 17:46:29 yeah, that is what I thought. 17:46:42 /* Send all of the headers now */ 17:46:42 ap_send_http_header(r); 17:46:59 The above for example. Just change it to the AOLserver call that sends the headers. 17:47:08 it has a part where it stores metadata about the files. I am not sure how that would work... 17:47:18 That'd be a cool project. 17:47:37 yeah, I wish I new C and had a month :) 17:48:42 mod_dav is fairly nicely written. I'm pleased. Not close to the mess that Apache is. 17:50:01 til (~tils@port-212-202-128-195.reverse.qsc.de) has joined #openacs 17:50:23 neat. the tricky part is all the improvments have gone into the apache 2.0 stuff. 17:50:39 I saw that. We'd have to grab that one. 17:50:48 davb: I wish I had a month :( 17:51:00 even a week :) 17:51:21 I don't think a week would be enough for this. It's a lot of code, then testing. 17:52:14 no, I just wish I had one. 17:54:47 here is cvs for the dav stuff in 2.0 http://cvs.apache.org/viewcvs.cgi/httpd-2.0/modules/dav/ 17:58:04 oh. Subversion is coming along nicely. 18:14:02 * rbm goes to class 19:13:33 davb has quit (Read error: 104 (Connection reset by peer)) 19:15:48 davb (~dave@rrcs-nys-24-97-22-203.biz.rr.com) has joined #openacs 19:30:32 hi davb 19:30:39 morning al 19:31:56 talilee has quit (Read error: 113 (No route to host)) 19:34:42 hi jim 19:39:34 ok, tried to delete package instance yesterday, from unmounted applications off of site map 19:42:09 talli_ (~talli@xcdfddb76.ip.ggn.net) has joined #openacs 19:42:24 talli_ is now known as talli 19:48:07 I got this: Query was not a statement returning rows.") 19:48:07 while executing 19:48:07 "ns_pg_bind 0or1row nsdb0 { 19:48:07 select case when count(*) = 0 then 0 else 1 end from apm_packages 19:48:07 where package_id = :package_id 19:48:07 }" 19:48:17 which looks ok. it should return 0 or 1 no matter what. 19:50:22 yeah, I got that too... but look at the associated .xqls 19:50:32 ah, I am looking back.. RI error 19:50:37 seems they put things in the wrong places? 19:50:39 centext index 19:50:45 context_index 19:51:10 wait, maybe that is because the other one failed. 19:51:13 let me look at the xql 19:54:44 does oracle support case? 19:54:51 dunno 19:54:56 til has quit (Read error: 113 (No route to host)) 19:54:59 but it does have decode 19:55:24 anyway, shouldn't matter for this, those queries are valid for postgresql. 19:55:51 so this problem is not actually this query? 19:56:29 what did you try to delete, btw 19:56:30 ? 19:56:44 unmounted wp instance. 19:56:58 I am not sure why it is saying query did not return any rows. 19:57:12 it looks ok. I copied the query frm the log and it returned 1 19:58:04 ah wait... 19:58:14 that query is run when there is an error deleting. 19:59:42 I think the error is in apm_package__delete 20:00:36 I don't see where apm_pacakge__delete will delete any objects created under the package. maybe acs_object__delete does that 20:02:55 yeah acs_object__delete should handle it. 20:03:15 I wonder if any other unmounted instances can be deleted. 20:03:21 of other packages that is. 20:04:04 dlk (dlk@as2-1-4.va.g.bonet.se) has joined #openacs 20:04:11 ok I mounted, unmounted and deleted a "page" application. 20:04:22 I know acs-subsite also has this problem. 20:05:27 wait, acs-subsite is a different bug I submitted it :) 20:08:14 ok the query that tries to see what is under the package_instance looks like it isn't going far enough down the tree maybe... 20:11:41 yep. that query is only catching the package object and the entry in apm_packages. not the presentation objects etc in the CR. 20:12:22 I suspect that breaks any package that has objects with the package_id ad context_id 20:12:31 :) 20:14:29 he davb and rbm, if you guys are interetsed in starting on mod_dav i'll try and scare up some money to finish it 20:15:18 I am only a $1/hour C hacker :) 20:15:34 probably about a millions hours though.... 20:15:37 yeah, but rob mayoff isn't :) 20:15:57 he would take considerably less I suspect. 20:16:04 hours that is. 20:16:30 yeah, but would cost considerably more 20:22:48 aha. it looks at the object_type hierarchy 20:23:28 so when you delete a package instance, that comes from acs_object so it knows to delete the apm_package and acs_object 20:23:49 I think there needs to be some sort of tcl callback for a package to delete or archive delete package_instance content. 20:33:18 talli has quit (Read error: 104 (Connection reset by peer)) 20:46:14 dlk has quit (Read error: 104 (Connection reset by peer)) 20:49:05 talilee (~talli@xcdfddb76.ip.ggn.net) has joined #openacs 20:49:29 talilee is now known as talli 20:50:10 davb, i'm torn 20:50:20 should i install debian or redhat? 20:51:21 anyone here wanna fight me about this/ 20:51:25 linux or BSD? 20:51:52 perhaps mandrake? 20:53:33 I like debian. 20:53:43 i think i'm going redhat 20:53:49 good reason? 20:56:12 you should install what works best for you... try a few different ones, see what you like/don't like 20:56:32 jim: ugh. i've installed enough OSes in my life. i want to relax 20:56:49 i'm doign a dual install right now - a win2K on one side of the office, redhat on the other 20:57:01 oh no 20:57:03 debian install is not so relaxing... but maintaining is easier 20:57:09 why don't you just install OS X and call it a career 20:57:21 because i don't have a mac yet! 20:57:24 the install is ok if you know the hardware :) 20:57:27 but i'm getting my G3 soon 20:57:33 redhat is probably the opposite 20:57:34 hold out for the g5 20:58:02 jim: rbm asked before, is any of the progeny installer going into debian? Would you even know? :) 20:58:04 davb: i'm having my OSX nerd friend resurrect a beige g3 i had given up for dead 20:58:06 easy install, maintainance not easy and standards compliance nil 20:58:15 talli: ah. cool. 20:58:19 jim: that's redhat? 20:58:38 well, the thing they're working on is something called debian-installer 20:59:14 what's current is something else called boot-floppies 20:59:37 yeah. I used the boot floppies for my server. 20:59:41 * jim must have coffee :) 20:59:47 bbiab 20:59:54 coffee.... good idea 21:01:06 ah... word is my G3 with OSX is happy. i'm glad 21:02:13 spiffy. 21:04:50 jim scared me... i'm going with debian 21:05:06 my experience with debian has been on an uncooperative laptop 21:05:14 this is on a desktop machine, so it should be easier 21:05:39 talli: "that's redhat?" well, exaggerating a little about the standards thing, but yeah, debian seems easy to maintain, if you follow certain rules 21:06:00 debian is super stable if you stay in stable also. 21:06:16 that was a terrible sentence. 21:06:29 and for an openacs, I -always- build everythnig 21:06:42 me too. 21:07:34 postgres in /usr/local/pgsql (and src in source/) and aolserver in /usr/local/aolserver (and src in source/) 21:08:05 except I installed the postgresql package on my web server, and I need to replace it with a compiled one so openfts will work. 21:09:06 rbm: I can't find your new debian maintainer application... so I couldn't advocate you yet... 21:09:23 (actually last time I looked was last nite) 21:09:41 I'll look again once coffee is in me :) 21:26:31 bye 21:26:34 davb has quit () 22:14:09 talilee (~talli@xcdfddb76.ip.ggn.net) has joined #openacs 22:14:09 talli has quit (Read error: 104 (Connection reset by peer)) 22:14:17 is shagster around? 22:14:31 talilee is now known as talli 22:14:47 jim: gave up on debian becuase my hardware is not known. 22:14:56 i'll have to suffer with redhat ;) 22:15:24 talli has quit (Client Quit) 22:17:40 talli (~talli@xcdfddb76.ip.ggn.net) has joined #openacs 22:19:12 talli: ok :) 22:22:28 that is a strength of redhat, it tries to identify hardware for you 22:22:39 that and suse 22:22:47 they both do a good job of that 22:24:15 damn :) "Foxy Lady" by hendrix as a somewhat funked up jazz tune 22:36:56 jim, check out sexmob.com 22:37:42 go to the goods section and listen to the "Nobody does it Better" mp3 22:42:17 hmm, I haev to set up my mp3 abilities... right now, I can jsut play them, not set up to browse/download/play 22:43:13 also, doesn't seem like those are mp3s... seems to be rm 22:50:26 moo 22:50:45 jim: I've used libdetect with Debian to autodetect hardware. It works just as well. 22:56:09 heya rbm... 22:56:16 didja apply yet? :) 22:56:28 jim: I was just going to... 22:56:32 cool 22:56:53 * rbm crosses his fingers, says a little prayer and heads to debian.org/devel/ again 22:57:25 rbm: best thing to do is apply and start trying to package simple stuff 22:57:38 jim: Will do so. 22:57:59 I want to package this Corel Visual Package Builder foo that I found at their soon-to-be-defunct opensource site 22:58:02 I'd like to see apm packages wrapped in debian packages 22:58:12 (to learn packaging) 22:58:19 jim: I want that too. 22:59:46 but initially, you could cvs checkout all the extra modules into a packages dir and package that 23:00:33 also, I was thinking you could have a pool that the packages could be unpacked in... 23:01:13 jim: Yesterday some folks at #debian-devel were saying they went through the entire nm process in 12 days! 23:01:30 yeah, others took longer... 23:01:41 look at rms' application :) 23:02:25 talilee (~talli@xcdfddb76.ip.ggn.net) has joined #openacs 23:03:07 talli has quit ("Trillian (http://www.ceruleanstudios.com)") 23:03:38 talilee is now known as talli 23:03:50 paje: where's shagster 23:03:50 talli: no idea 23:03:56 paje: shagster 23:03:56 talli: what? 23:04:04 paje: last shagster 23:04:04 talli: excuse me? 23:04:09 paje: i hate you 23:04:09 talli: excuse me? 23:04:19 paje: seen shagster? 23:04:19 shagster was last seen on #openacs 1 days, 18 hours, 27 minutes and 55 seconds ago, saying: Hmmm...debian getting a new developer? :) [Sat Feb 23 21:38:19 2002] 23:04:25 thanks rbm 23:04:27 talli: :) 23:08:59 ok, lessee... 23:10:33 you could have the base of openacs in a package, installing this package loads it into, say, /usr/lib/openacs-base 23:10:52 then you could have an instance creator 23:11:32 which copies the whole tree into /web/servername 23:12:28 then, each package .deb installs its file content into say /usr/lib/openacs-pkgs 23:13:03 what do you free software guys use for word docs, openoffice, abiword or koffice? 23:13:17 talli: to view or to create? 23:13:21 both 23:13:50 it would really fscking suck if i had to use two editors, one to view one to create 23:13:55 to quickly view (remember my system is old) I use antiword. To create I usually use abiword. 23:13:55 and the instance creator or something else could copy selected package trees into /web/servername/packages 23:14:08 i just tried to open a pretty complex word doc and kword couldn't deal 23:14:20 antiword, huh? sounds cool 23:14:34 talli: For complex word docs, openoffice is the best. abiword will crash. 23:14:43 ah, ok 23:14:45 talli: antiword converts to other formats, HTML, PDF, et. 23:15:03 if you copy the files every time, then deleting the whole pkg won't break the actual deb install 23:15:04 i don't trust openoffice though, as its sun. i fear it's close to "shared source" 23:15:16 (sorry jim for interrupting your train) 23:15:28 jim: hmmm. 23:15:45 interrupting trains is done at your own risk :) especially when they are moving :) 23:16:14 is the "jim on coffee" train heavy and painful when encountered? :) 23:16:38 ohh, you mean train of thought :) 23:18:48 that didn't get interrupted badly :) 23:18:50 do i? i thought i meant BART 23:19:11 Being Away Really Tingles/ 23:19:12 ? 23:19:56 uh, more or less 23:20:49 I got the idea out there.... rbm: I'm thinking of ways you might package it such that you don't have to add things to the openacs itself in order to support it being enclosed in a package 23:21:45 jim: I appreciate any suggestions. 23:22:07 here's the scenario I was thinking of just then... 23:22:23 have base of openacs installed, data model loaded. 23:22:54 install a package from a .deb, 23:23:07 installing just the base won't be easy currently 23:23:07 then add that package to the openacs instance 23:24:02 by the base I mean everything you get when you check out openacs 4 23:24:21 (kernel, subsite, skin, workflow, cr, etc) 23:24:42 Hmmm. That's a bit more than just "base" don't you think? 23:25:16 ok, I'm using the wrong word... 23:25:30 can we overload the word base for the moment? 23:25:47 jim: Sure :-) 23:26:01 so here's what I'm thinking... 23:26:19 you install the base, 23:26:30 then you "make an instance" 23:27:26 making the instance (let's say this instance is called fooacs) copies all the base files to /web/fooacs/ 23:30:41 it also writes out an aolserver config file and starts an aolserver, and tells you to go to the url and do the install (hitting the next button a coupla times, filling in the make-admin and site-info forms) 23:31:05 at which point the aolserver finishes and terminates just like it usually does 23:31:58 so now you have the data model installed and the base and config file in place 23:32:45 the script could start that instance, or tell you how 23:34:32 so, then the instance somehow gets started 23:35:34 and now you want to install a package 23:37:02 probably the simplest approach that won't have mutual breakage between debian and openacs is to have a place for the packages so they can be copied into the instance tree, possibly multiple times 23:37:53 (allowing here for the possiblity the admin deletes the package from his site entirely: doing so should not delete the files from the "staging area" 23:37:56 ) 23:38:32 so, putting the package into the instance is as simple as copying a file tree 23:39:23 or untarring, if you want to store the base staging and package staging as compressed tarballs 23:40:13 davb (dave@alb-24-58-162-46.nycap.rr.com) has joined #openacs 23:40:25 heya davb 23:40:32 hi 23:41:14 davb: about this instance-delete thing... 23:42:56 instance-delete.xql has: 23:43:03 package_mounted_p 23:43:13 instance_delete_doubleclick_ck 23:43:29 instance-delete-oracle.xql has: 23:43:45 package_instance_delete 23:44:06 instance-delete-postgresql.xql has: 23:44:08 package_instance_delete 23:45:17 should instance_delete_doubleclick_ck be moved into -oracle and -postgresql? 23:45:25 jim: ok. My opinion is that the package deletion is inherently broken. it does not handle a case where any other acs_objects refer back to the pacakge_id. THat is any items a package created are not removed by pacakge_instance__delete 23:45:36 jim: I think so, it has the "case"? 23:45:42 yes 23:45:54 then yes case for postgresql, decode for oracle. 23:46:49 so apm_package.delete is broke because it doesn't get rid of data associated with the package 23:46:52 I wonder what the docs have to say about removing a package instance. I am suspecting the apm people hoped the package people would be rid of their own content before trying to delete an instance. 23:47:27 jim: yes. or rather the data is still there when it is called. I am not sure it should necessarily delete the content itself (apm_pacakage.delete that is) 23:49:19 ok, say (general case: any package) admin decides to delete a package... if package is later reinstalled, how could you access the data? 23:51:13 good question. I am just not sure how the original authors intended it to work. of course, there is a good chance they didn't think of it. 23:51:46 I think maybe there should be a tcl callback or somethign registered with the package so when you try to delete it, it goes and removes all its stuff. 23:52:06 no matter how you look at it, deleting of a package instance is an extreme action 23:52:17 right. 23:52:46 I am just not sure we can generalize the removal of package content enough... maybe we can if they use the acs_object system, but you are not required to use it ... 23:52:58 * davb reads some docs 23:53:06 where are you reading? 23:55:42 subsite, apm... 23:56:54 "10.20.0 Package deletion 23:56:54 The system must provide an API call to delete a package and all related objects in the subsite's context. 23:56:54 " 23:57:04 in subsite requirements doc. 23:57:10 I guess they didn't finish. 23:57:36 It seems like this should have been covered. I would hate to duplicate work, but it looks like aD never got around to it. 23:58:28 I can't find a doc that mentions multiple instances of a package, etc...