Menu
▾
▴
wpigdc-cms-development — This is a list for all interested parties concerning the backend development
You can subscribe to this list here.
| 2001 |
Jan
|
Feb
|
Mar
|
Apr
|
May
|
Jun
|
Jul
(20) |
Aug
(26) |
Sep
|
Oct
|
Nov
|
Dec
|
|---|
|
From: Dan P. <dp...@WP...> - 2001-08-17 13:22:18
|
On Fri, 17 Aug 2001, Liam Morley wrote: > As you'll see in the source as well as on www.wpi.edu/~lmorley/puppets, > there have been changes to the News section (I've uploaded screenshots) so > that it can be configured to the individual requesting page. A lot of it > is kludgy, however I ask that you don't fix any kludges until after beta > unless it's pretty urgent and the rest of the site is done and tested. The next time *anyone* get's the urge to commit "kludges" that'll be changed in CVS - PLEASE CREATE A BRANCH IN CVS! You're changes are completely counter to the eventual design of the NewsSource and will be removed - I'm probably going to just revert to the version before your change. If you don't know how to create branches, please read the manual: http://www.cvshome.org/docs/manual/cvs_5.html That way the working CMS code can stay up-to-date and the other code can continue without effecting development. > Dan, I need to know the status of user-session management. Please let me > know everything I need to get set up as well as have others test. Please elaborate - this gives me absolutely to work with. "It works?" I dunno - it seems to work for me? Because of serious security flaws I'm not going to allow any actual posting via the website until they are revised sometime next week? > > I'd also appreciate it if you can set up an account for me on GDC as soon > as you get it up (ie, before you install, so I can help configure). I'd > like to check on the configuration before it goes up. You can make up a > password for me beforehand and send it to me w/PGP, then create the user > once the machine is up. Please let me know if you'll be able to do that... > That'd be really great. > This bit should probably either have A) been sent personally to me or B) via a WPI list. There are *publically available* archives of this list and this list should be kept to simply development issues - and *not* GDC webpage issues. The CMS backend is the backend we're using - it is *not* the GDC webpage. The GDC webpage is the setup we are using on gdc.wpi.edu. |
|
From: Dan P. <dp...@WP...> - 2001-08-17 13:11:31
|
I'd rather not break SAX, thanks anyway... Classes implementing XMLConsumer receives the SAX events - the class receiving them will "know" how to handle them. Since the class we're sending SAX events is usually TraxTransformer, there's no way for us to extend the class. You might be able to add the functionality to a custom DataSource extension. An XMLConsumer is a fancy name for an org.xml.sax.DocumentHandler and a org.xml.sax.ext.LexicalHandler - the two interfaces needed to completely grok an XML document. That's it - it's an interface. You can't "extend" it anyway. On Fri, 17 Aug 2001, Liam Morley wrote: > I didn't see any ability within XMLConsumer to determine what element you > are working within. If that's the case, I think we should subclass to > extend this functionality. If that's not the case, please let me know. |
|
From: Liam M. <lm...@WP...> - 2001-08-17 06:46:41
|
As you'll see in the source as well as on www.wpi.edu/~lmorley/puppets, there have been changes to the News section (I've uploaded screenshots) so that it can be configured to the individual requesting page. A lot of it is kludgy, however I ask that you don't fix any kludges until after beta unless it's pretty urgent and the rest of the site is done and tested. I need to stop staying up until 2:30AM to make sure this site gets off the ground on time, my boss is getting upset... I'll create the graphics for the main sections of the site tomorrow or Saturday, and upload those to the puppets directory as well... Dan, I need to know the status of user-session management. Please let me know everything I need to get set up as well as have others test. I'd also appreciate it if you can set up an account for me on GDC as soon as you get it up (ie, before you install, so I can help configure). I'd like to check on the configuration before it goes up. You can make up a password for me beforehand and send it to me w/PGP, then create the user once the machine is up. Please let me know if you'll be able to do that... That'd be really great. If anybody needs any more info, please let me know. I'm off to bed... ------- Liam Morley "light the deep, and bring silence to the world. light the world, and bring depth to the silence." |
|
From: Liam M. <lm...@WP...> - 2001-08-17 04:09:39
|
I didn't see any ability within XMLConsumer to determine what element you are working within. If that's the case, I think we should subclass to extend this functionality. If that's not the case, please let me know. ------- Liam Morley "light the deep, and bring silence to the world. light the world, and bring depth to the silence." |
|
From: Liam M. <lm...@WP...> - 2001-08-17 03:14:43
|
There was a malformed javadoc tag. It's been fixed, and the changes are reflected in CVS. ------- Liam Morley "light the deep, and bring silence to the world. light the world, and bring depth to the silence." |
|
From: Dan P. <dp...@WP...> - 2001-08-17 01:41:03
|
That's been fixed in the latest CVS, as well as a bugfix to the current NewsSource output (the new output contains ALL the data a NewsStory contains). The original change involved linking the User who authored the story from the NewsStory object instead of just having the numeric ID - this breaks the GUI anyway, but it'll compile at least. Last but not least, it won't hurt to keep the GUI disabled anyway - it's mostly used for testing the DB stuff right now (although eventually it'll probably be made into a full-fledged tool). On Thu, 16 Aug 2001, Liam Morley wrote: > After updating today, I found that there was a build error in the GUI > code. Since the GUI code isn't necessary for CMS (ie it isn't used by > Cocoon), I simply excluded it from the build by editing my build.xml file. > If you'd like to know how to do this, you can do so as follows: > > 1. open build.xml, which is available in CVS > 2. search for <target name="prepare-src" .... > 3. within the fileset tag (right around where it says 'no excludes yet'), > you want to put in the following line: > > <exclude name="**/gui"/> > > That should do it. I should note that this is a TEMPORARY FIX; if we ever > need to use the GUI in the future, you'll have to take this line out. > However if we don't need the gui, you can keep the exclude in there to > make your build go faster. > > I'm going to be integrating the news section into the front-end of the > website, so I'll probably be editing this code. Let me know if any of you > are working on it as well so we can successfully merge without too much > of a hassle (I prefer not to use the auto-merge in CVS, I consider that > to be the definition of "too much of a hassle"). |
|
From: Liam M. <lm...@WP...> - 2001-08-17 01:14:04
|
After updating today, I found that there was a build error in the GUI code. Since the GUI code isn't necessary for CMS (ie it isn't used by Cocoon), I simply excluded it from the build by editing my build.xml file. If you'd like to know how to do this, you can do so as follows: 1. open build.xml, which is available in CVS 2. search for <target name="prepare-src" .... 3. within the fileset tag (right around where it says 'no excludes yet'), you want to put in the following line: <exclude name="**/gui"/> That should do it. I should note that this is a TEMPORARY FIX; if we ever need to use the GUI in the future, you'll have to take this line out. However if we don't need the gui, you can keep the exclude in there to make your build go faster. I'm going to be integrating the news section into the front-end of the website, so I'll probably be editing this code. Let me know if any of you are working on it as well so we can successfully merge without too much of a hassle (I prefer not to use the auto-merge in CVS, I consider that to be the definition of "too much of a hassle"). ------- Liam Morley "light the deep, and bring silence to the world. light the world, and bring depth to the silence." |
|
From: Liam M. <lm...@WP...> - 2001-08-13 22:35:29
|
Thanks, Oleg, that's right. I had thought about that earlier (what happens when the title is "General Assembly" or something..), which narrows it down. ------- Liam Morley "light the deep, and bring silence to the world. light the world, and bring depth to the silence." |
|
From: Oleg R. <sv....@ve...> - 2001-08-13 12:21:25
|
On Monday, August 13, 2001, 1:41:55 AM, in message "[Wpigdc-cms-development] how to customize the title dependent on the page", Liam wrote: LM> general help question. I want events/past.html to have a different <TITLE> LM> than events/future.html . XML docs and XSLT docs don't know the actual URL LM> that the user is at, so I can't do it there. Only two places that I know LM> of actually have this information; the sitemap and the EventSource class. LM> So.. either it's passed in as a parameter to the XML doc in sitemap, or LM> its sent as a SAX event in EventSource. I'm not sure if either are LM> possible, and if either one is, I don't know how to do it. I've made a few LM> guess stabs, but nothing works. Any ideas? Placing the title in the sitemap looks like a bad idea to me--what if you need to customize the title of the page according to the data fetched from the database? |
|
From: Liam M. <lm...@WP...> - 2001-08-13 05:41:59
|
general help question. I want events/past.html to have a different <TITLE> than events/future.html . XML docs and XSLT docs don't know the actual URL that the user is at, so I can't do it there. Only two places that I know of actually have this information; the sitemap and the EventSource class. So.. either it's passed in as a parameter to the XML doc in sitemap, or its sent as a SAX event in EventSource. I'm not sure if either are possible, and if either one is, I don't know how to do it. I've made a few guess stabs, but nothing works. Any ideas? ------- Liam Morley "light the deep, and bring silence to the world. light the world, and bring depth to the silence." |
|
From: Dan P. <dp...@WP...> - 2001-08-12 04:20:52
|
On Sun, 12 Aug 2001, Liam Morley wrote: > Let me reiterate that NO ONE- not Dan, not I, no one; can alter an agreed > upon agenda by themselves unless both Dan and I agree to it. This is the > current situation, and after beta, I'll be in full control. So events will > still be part of alpha 1, whenever alpha 1 does get released. Alpha 1 is > not considered a marked deadline until basic event functionality is reached. WPIGDC-CMS is the BACKEND and is the open source code. The GDC webpage is something *entirely* different. It's probable that Alpha 2 won't be released until sometime well after the GDC site hits beta. The first CMS beta probably will not be released until well after September. |
|
From: Liam M. <lm...@WP...> - 2001-08-12 04:10:23
|
Dan, you are not available often enough for me to ask you first, so I can not do that. Format is not what's important for these documents at the moment, as they are not linked from anywhere. What's much more important is that everybody is brought up to speed so we can get more people involved. This is a high priority. As I mentioned previously, these documents are in disarray, which will change when we're back on campus. Let me reiterate that NO ONE- not Dan, not I, no one; can alter an agreed upon agenda by themselves unless both Dan and I agree to it. This is the current situation, and after beta, I'll be in full control. So events will still be part of alpha 1, whenever alpha 1 does get released. Alpha 1 is not considered a marked deadline until basic event functionality is reached. ------- Liam Morley "light the deep, and bring silence to the world. light the world, and bring depth to the silence." |
|
From: Dan P. <dp...@WP...> - 2001-08-12 03:39:55
|
Would you please send stuff through me before posting anything to the CMS webspace? There's a place for these things and a style that the main web page uses. (Which will be changing when I release Alpha 1, so don't worry about it quite yet - but everything that's up there WILL be deleted when the new stuff goes live. On that note, please remove the badly out-of-date Java API docs - they're just wasting space and new ones are available via http://wpigdc-cms.sourceforge.net/docs/javadoc/ which are kept (relatively) up-to-date.) On a completely different note, CMS Alpha 1 hopefully will be ready tomorrow, but I need to go over the event implementation and probably remove the pairs thing and add a method to DataSource that passes XHTML through as SAX events. (It's also possible I'll just leave events out of the official CMS Alpha 1.) (NOTE: If you ever get the urge to implement a hash map in an RDBS again, please use two tables: one for the main data, one for the pairs. Like so: CREATE TABLE cms_event_pairs ( cms_event_id INTEGER NOT NULL, cms_key VARCHAR(64) NOT NULL, cms_value TEXT, PRIMARY KEY (cms_event_id, cms_key)); Alright, I lied - MySQL isn't a RDBS, but it works well enough as one for that to work.) |
|
From: Dan P. <dp...@WP...> - 2001-08-12 03:28:50
|
A couple of things which'll need to be pointed out... 1) Every CMS source class MUST start with a JavaDoc comment, which MUST include the following tags: /** * @author $Author$ * @version $Revision$ */ This allows auto-tagging of the data via CVS. 2) Please inform me before commiting ANYTHING to CVS that relates to the DB component - I had to merge your changes back with mine and am likely going to be changing portions based on new code. (Also, do NOT print stack traces in the DB engine - it's now possible to embed Throwable's into a DatabaseException, so do that instead. Plus DBIterator is probably going to be extended to offer error catching services.) 3) Do NOT catch SAXExceptions in your DataSource - there's absolutely NO need. If a SAXException is thrown, then it's likely that the entire response is screwed ANYWAY, and Cocoon should be allowed to handle it normally. On Sat, 11 Aug 2001, Liam Morley wrote: > I removed EventSource from the > repository, but you can still see it there in the "Attic". You're supposed to - the file were in the repository and can be retrieved - just like any version of any file. I hope to be tagging the repository as Alpha 1 soon, but the events will probably delay that a week. > by the way: as far as events go. Dates haven't really been worked out. > Currently two sets of dates are sent through SAX: > start-date-long/end-date-long and start-date-short/end-date-short (the > long versions will give you "March 13th, 2001" whereas the short versions > will give you "03/13/2001"- the methods I wrote to do this might become > handy utility methods, in which case eventually we may write a lil utils > package and throw stuff like this there). Have you ever checked java.text.SimpleDateFormat? I think you were wasting time... (Speaking of which, it should be simple to change to allow the source XML to specify the SimpleDateFormat format...) > I made a few small changes to the schema I set up, largely adding a unique > string field for URL detection. Here's how that works. Everything in > either "events", "events/", or "events/*.html" is sent to "events.xml". > Then, the datasource figures out what the url is. if it's > past.html, it serves up all past events (to be changed when user > preferences is firmed up and users can select how many past events > they want displayed); if the URL is future.html, it serves up > future events; if the URL ends with events, events/, or events/index.html, > it serves up a past and future (just like on the index page as it's > currently implemented). Anything else goes to the event described in the > URL. for example: > > /events/gen_assembly_033101.html > > EventSource will look in the database for an event with a "cms_short_name" > of 'gen_assembly_033101'. If it finds it, it serves the information > associated with it; if it doesn't, it handles it as a 404. Currently 404 > stuff is handled directly within the EventSource itself (the 404 is caused > by an exception in the result set, I can probably throw whatever I want > though in the future when we want to handle stuff like this universally; > as we're not ready yet, it's going in EventSource). Keep in mind this is a > virtual 404; it's not actually searching for a gen_assembly_033101.html. > Such is the miracle of Cocoon... COMPLETELY transparent to the user. I'm not sure how 404 processing will work... we'll see... (otherwise this sounds like a good solution, but I'm not sure I like the "short event name" idea... whatever. |
|
From: Liam M. <lm...@WP...> - 2001-08-12 03:26:35
|
As promised, I finally have finished my in-depth report on installing server software on a win32 platform. The setup was done in Win98, however Win2000 wasn't hard to set up either. A complete guide to downloading, installing, and running/testing Apache, Tomcat, and Cocoon can be found at: http://wpigdc-cms.sourceforge.net/install-win32.html This leaves those who haven't installed without an excuse :o) Please, all of you who are installing, let me know how it goes. If it doesn't go smoothly, let me know. If you notice, the webdev resources are largely unorganized. This will likely remain to be the case until after school starts, when we have more people than just Dan and I doing the core work. ------- Liam Morley "light the deep, and bring silence to the world. light the world, and bring depth to the silence." |
|
From: Liam M. <lm...@WP...> - 2001-08-11 20:55:15
|
I love cygwin. I love vi. I HATE cygwin vi. You'll see some nasty log entries for a few files; if you were looking over my shoulder while I was trying to use this bastardized editor, you'd understand why. You'd sympathize. It refreshes the screen when it wants to, seemingly completely at random. You have no idea where the cursor is, except for a little number at the bottom telling you row-column, and when it specifies two columns simultaneously you know you're fucked. Also, when I started writing the events classes, sometimes I copied directory structures. Like I copied sources/news to sources/events. However, forgetting to delete the CVS directory underneath events, EventSource got checked in to the news directory. ARRGH. It was a happy check-in. Ok, it went miserably. I removed EventSource from the repository, but you can still see it there in the "Attic". Never. Never. EVER. NEVER EVER EVER EVER NEVER. Try the cygwin vi editor. Those of you who don't like vi... you'd cry. Those of you who love vi (like me)... you'd cry. you really would. you really really really would. Events are added to CVS... however the XSLT stuff for it is all on my HD, as XML/XSL has yet to hit CVS. So, if you want to see a working version, write your own XSL :o) no, I'm kidding. ha ha ha. I'll send you along what I have, just email me if you want it. by the way: as far as events go. Dates haven't really been worked out. Currently two sets of dates are sent through SAX: start-date-long/end-date-long and start-date-short/end-date-short (the long versions will give you "March 13th, 2001" whereas the short versions will give you "03/13/2001"- the methods I wrote to do this might become handy utility methods, in which case eventually we may write a lil utils package and throw stuff like this there). There is currently no support for time as I'm not sure how the time will be stretched, but to tell you the truth if time is left out until after beta I won't be upset. Time IS supported in the database however, as the schema is using DATETIME. I made a few small changes to the schema I set up, largely adding a unique string field for URL detection. Here's how that works. Everything in either "events", "events/", or "events/*.html" is sent to "events.xml". Then, the datasource figures out what the url is. if it's past.html, it serves up all past events (to be changed when user preferences is firmed up and users can select how many past events they want displayed); if the URL is future.html, it serves up future events; if the URL ends with events, events/, or events/index.html, it serves up a past and future (just like on the index page as it's currently implemented). Anything else goes to the event described in the URL. for example: /events/gen_assembly_033101.html EventSource will look in the database for an event with a "cms_short_name" of 'gen_assembly_033101'. If it finds it, it serves the information associated with it; if it doesn't, it handles it as a 404. Currently 404 stuff is handled directly within the EventSource itself (the 404 is caused by an exception in the result set, I can probably throw whatever I want though in the future when we want to handle stuff like this universally; as we're not ready yet, it's going in EventSource). Keep in mind this is a virtual 404; it's not actually searching for a gen_assembly_033101.html. Such is the miracle of Cocoon... COMPLETELY transparent to the user. (note to self: install a REAL vi editor... scratch that, install a REAL Operating System.) Of course with the way that cvs is installed on this machine, merges are somewhat difficult, so please make sure nothing got thrown out. I apologize profusely if I fucked anything up. I can't wait for my CD-RW, processor, and mobo to arrive so I can install anything I want... ------- Liam Morley "light the deep, and bring silence to the world. light the world, and bring depth to the silence." |
|
From: Liam M. <lm...@WP...> - 2001-08-09 16:38:06
|
With Oleg's help, I found the following: select cms_event_group.*, min(cms_start_date) from cms_event_group, cms_event where cms_event.cms_event_group = cms_event_group.cms_id GROUP BY cms_event_group.cms_id; This seems to get the job done. If anybody sees a problem in this, let me know. ------- Liam Morley "light the deep, and bring silence to the world. light the world, and bring depth to the silence." |
|
From: Oleg R. <sv....@ve...> - 2001-08-09 15:26:11
|
(using a different e-mail address in the From: field, if you reply to
that address, I will get very upset).
On Thursday, August 09, 2001, 4:21:00 AM, in message
"[Wpigdc-cms-development] MySQL doesn't support subqueries", Liam
wrote:
LM> OK. So. It's 3:45 AM, so go easy on me.
LM> Here's what I want to do. We want to sort events by their start
LM> date. Unfortunately, events themselves don't have a start date in
LM> the db; their sub-items (or whatever you want to call it) do.
LM> (Read the previous email to understand). This wouldn't be a
LM> problem, if MySQL supported complex joins. But it doesn't. Just
LM> the simple stuff.
MySQL supports pretty damn complex joins, it does not support
sub-selects.
LM> Here's what I want to do: I want to select the earliest date item
LM> from the cms_event table where the cms_event_group foreign key
LM> corresponds to that particular event group, like so:
LM> select start_date from event_item where event_id = [main_event_id]
LM> order by start_date limit 1;
LM> (if there's a better way, please let me know.)
SELECT MIN(start_date) FROM event_item WHERE event_id = [main_event_id]
LM> then, I want to tack that start date onto the event_group table so
LM> I can sort by it. And I'd like to do that in 1 SQL statement. More
LM> complex databases will let you do that. However, it doesn't seem
LM> like MySQL supports something like that (PLEASE correct me if I'm
LM> wrong, and show me the proper syntax).
How about:
SELECT cms_event_group.cms_id, cms_event_group.cms_name,
...[as many fields from cms_event_group as you need]...,
MIN(cms_event.cms_start_date)
FROM cms_event_group, cms_event
WHERE cms_event_group.cms_id = cms_event.cms_event_group
Let me know if that works...
LM> So. A few choices. d. every time we want to sort by start date for
LM> the event group (which is the most common field for sorting),
LM> create a new temporary table that does the same as the join. Then,
LM> when finished, delete the table. This is expensive, and if
LM> something goes wrong and the tables don't get deleted, wasteful.
LM> People will start DOS'ing our ass simply by hitting "Refresh" on
LM> the events page. Ok, I don't know if it's that expensive, maybe
LM> it's worth finding out.
MySQL has a few choices for temporary tables that are automatically
deleted at the end of the session (that's the definition of
"temporary"), including a "memory" table that doesn't even get written
to disk.
LM> e. Forget sorting using the database: sort in the object model.
LM> Certainly possible, but it doesn't work with the current setup
LM> using DBIterator at all. It's also most likely much slower than
LM> sorting in the db.
Ugh, won't touch the issue....
|
|
From: Liam M. <lm...@WP...> - 2001-08-09 08:21:03
|
OK. So. It's 3:45 AM, so go easy on me. Here's what I want to do. We want to sort events by their start date. Unfortunately, events themselves don't have a start date in the db; their sub-items (or whatever you want to call it) do. (Read the previous email to understand). This wouldn't be a problem, if MySQL supported complex joins. But it doesn't. Just the simple stuff. Here's what I want to do: I want to select the earliest date item from the cms_event table where the cms_event_group foreign key corresponds to that particular event group, like so: select start_date from event_item where event_id = [main_event_id] order by start_date limit 1; (if there's a better way, please let me know.) then, I want to tack that start date onto the event_group table so I can sort by it. And I'd like to do that in 1 SQL statement. More complex databases will let you do that. However, it doesn't seem like MySQL supports something like that (PLEASE correct me if I'm wrong, and show me the proper syntax). So. A few choices. a. throw MySQL out, get a new database ;o) might be best in long term, not great in short term. b. include the start date in the cms_event_group table (redundant! theoretically bad! every time you update one, you have to update the other!) c. rework the events section of the schema in some other way that is not redundant and also works in MySQL. d. every time we want to sort by start date for the event group (which is the most common field for sorting), create a new temporary table that does the same as the join. Then, when finished, delete the table. This is expensive, and if something goes wrong and the tables don't get deleted, wasteful. People will start DOS'ing our ass simply by hitting "Refresh" on the events page. Ok, I don't know if it's that expensive, maybe it's worth finding out. e. Forget sorting using the database: sort in the object model. Certainly possible, but it doesn't work with the current setup using DBIterator at all. It's also most likely much slower than sorting in the db. thoughts? favorite option? other ideas? ------- Liam Morley "light the deep, and bring silence to the world. light the world, and bring depth to the silence." |
|
From: Liam M. <lm...@WP...> - 2001-08-09 04:27:21
|
Theoretically, there shouldn't be a reason to include db specific information (like numerical id's) in a Java object model; it should be transparent at that level. So my question is, why do we have them? Can we get rid of them? What was the original intention when putting them in? ------- Liam Morley "light the deep, and bring silence to the world. light the world, and bring depth to the silence." |
|
From: Liam M. <lm...@WP...> - 2001-08-09 01:50:38
|
Doing preliminary work on the events section. Here are the tables I've created (following the format Dan outlined in his schema): CREATE TABLE cms_event_group ( cms_id INTEGER PRIMARY KEY, cms_name VARCHAR(64) NOT NULL, cms_location VARCHAR(128) NOT NULL, cms_url VARCHAR(255), cms_pairs TEXT ); cms_pairs will be a string containing name/value pairs for those things to be included for that event that we either haven't thought of or don't apply to events in general. I'll explain the name "event_group" below. CREATE TABLE cms_event ( cms_id INTEGER PRIMARY KEY, cms_start_date DATETIME NOT NULL, cms_end_date DATETIME NOT NULL, cms_pairs TEXT, cms_event_group INTEGER ); Here is where the start date and end date are kept. As far as event_group vs. event goes: I'll use an example. Say you've got an event (let's call it "Kill 'em All 2001") spanning 2 days. The first day, it starts at 6pm and lasts until 2am. The second day, it starts at 11am and lasts until 3pm. You want seperate information for each start/end date. So you create 2 cms_event records and one cms_event_group record, with the 2 cms_events linked to the one cms_event_group. Of course this could go into a name/value pair within cms_event_group, but I want to keep name/value pair usage to a minimum (and if there's something that's getting placed in name/value pairs often, I'd like to move it up and create an actual attribute for it). What I didn't include in here is any link to future work done in ACL's or forums. These won't be in the database at all until after beta, and I'm following Dan's example by keeping the db clean until it needs to get dirtied. We can stage different schema versions over time without much of a problem; this would be the desired approach in my eyes. Also, I didn't include cms_groups either, although that does look applicable. Dan, what's the purpose of cms_groups? You've got news linked to it, is that something you intended to link news articles, events, projects, and forums? Comments are more than welcome. Quick question: are there plans for ACL's for the beta? Or are we starting out with all events/news articles accessible by all then going from there? ------- Liam Morley "light the deep, and bring silence to the world. light the world, and bring depth to the silence." |
|
From: Liam M. <lm...@WP...> - 2001-08-06 20:16:01
|
Yep, I noticed :o) first.. we don't use "CVS" per se at EMC, but we use a larger more robust system (with plenty of hassles nonetheless) called "ClearCase". It's proprietary. We do have a lib folder in our development branches, and in this folder are included jar files from 3rd party software we are licensing (for example, classes for DB Mapping 3rd party software, classes for Java GUI 3rd party software, as well as the oracle drivers for database connectivity). I'm planning on checking out Cocoon's source from CVS to see how they do it. If they include 3rd party stuff, I see no reason why we shouldn't. If they don't, well then we don't have to either :o) but it's not a bad idea.. that's what the repository is there for. 2nd, I've already just about finished a batch file for Ant building, copied almost word for word from Cocoon's batch file (except modified to include relevant jars). It requires a %TOMCAT_HOME% env variable (it expects that you're putting the jar files in your classpath dynamically by adding them to the tomcat folder), as well as a lib folder that houses ant and ant-optional. If we do decide to package jars with CVS, then the TOMCAT_HOME requirement becomes unnecessary. Of course adding duplicated files to CVS does increase the size of the tarball, but I'm not too worried - yet. 3rd, I've thought about creating an installation kit, but I don't think it should be any more robust than Cocoon's installation "kit" (however, we better have much better documentation). We certainly don't need an executable like Apache's installation. I'll start working on something as soon as the heirarchy of the CVS tree has become closer to finished. ------- Liam Morley "light the deep, and bring silence to the world. light the world, and bring depth to the silence." |
|
From: Dan P. <dp...@WP...> - 2001-08-06 15:41:38
|
On Mon, 6 Aug 2001, Dan Potter wrote: > You don't add binaries into CVS - more on this later though. Who else noticed that I never got back to this? Anyhoo, here's the "more:" Sometime this week, I hope to package up the current system into a binary distribution and post it on Sourceforge as "Alpha 1." This distribution would include: 1) Tomcat 3.2.3 2) Cocoon 2 3) The MM-MySQL JDBC driver 4) The CMS "stuff" 5) And I'd like to include an installer - but am not going to waste time on building one. The various component software would be set up to use the CMS system "out of the box" so to speak. This is something I hope to do soon, but unfortunately, it's unlikely to actually get done this week (as things currently stand, there's enough info to get people online - I'm going to add the MySQL driver bit to the install documentation, it's sorta-there in the "Install MySQL" bit, but that's not really good enough). Mainly because I hope to be working on more important things, like allowing posting of news stories through the website (although the first step is actually fixing up the RemoteConfig utility to be able to post stories and add users and the like). |
|
From: Dan P. <dp...@WP...> - 2001-08-06 13:51:41
|
Replies below. On Mon, 6 Aug 2001, Liam Morley wrote: > I was able to get ant to work without too much problem, but I think we > need to include batch/shell scripts in CVS to make it easier on those who > don't know what ant is (which, as far as I know, is most if not all people > in the webdev team other than you and I). This usually is added later in the dev cycle and as part of ZIP and tarball distributions - it wouldn't hurt to write it out now, except that it would require us to include ant. This is probably something that best waits until we write up a source distribution. > We should also add the mysql.jar file in CVS as well I think, or at least > give a link to where it can be found. People without JDBC experience will > get 'invalid driver' exceptions without knowing what to do. You don't add binaries into CVS - more on this later though. > Also, I noticed the SQL script- that's great. I created a DB on my home PC > (as well as work PC) and was able to get everything to run nicely using > the RemoteConfig class. What I didn't see in CVS or the SF site is the > XML code; I was wondering if I'm looking in the wrong places, or are you > waiting to post that? I started creating "default" XML pages on Saturday, and will probably be posting them to CVS ... sometime this week. Along with these are the default XSL pages and CSS - it isn't very "pretty," but the idea is to make people customize the system :) > As far as RemoteConfig goes.. looks good so far. Maybe we can use this for > officers to add things to the database as well (taking their > usernames/passwords out of the conf file, we'll know what db rights to > grant them). Or we go with the web-interface idea.. but both are a > thought. Eventually, the RemoteConfig program will be capable of initializing the database - the actual code for setting up the tables is supposed to be the DBEngine's responsibility, which is why the script isn't part of the CVS repository. (Actually, the idea is that the system will initialize the database the first time it's started.) The bottom line is that the database should be "engine independent" and that each DBEngine has the freedom of creating the database however it likes. The RemoteConfig program currently isn't capable of posting stuff, it's main use right now is testing the DBEngine :) (Likewise, the overuse of System.out in the classes are more for testing the way it works and will probably be removed permantently by the first beta.) |
|
From: Liam M. <lm...@WP...> - 2001-08-06 08:02:56
|
Dan, this is great. Just a few quick comments, though. I'm sure you're already addressing some of these, but I just wanted to make sure. I was able to get ant to work without too much problem, but I think we need to include batch/shell scripts in CVS to make it easier on those who don't know what ant is (which, as far as I know, is most if not all people in the webdev team other than you and I). We should also add the mysql.jar file in CVS as well I think, or at least give a link to where it can be found. People without JDBC experience will get 'invalid driver' exceptions without knowing what to do. Also, I noticed the SQL script- that's great. I created a DB on my home PC (as well as work PC) and was able to get everything to run nicely using the RemoteConfig class. What I didn't see in CVS or the SF site is the XML code; I was wondering if I'm looking in the wrong places, or are you waiting to post that? As far as RemoteConfig goes.. looks good so far. Maybe we can use this for officers to add things to the database as well (taking their usernames/passwords out of the conf file, we'll know what db rights to grant them). Or we go with the web-interface idea.. but both are a thought. Again if you've already thought of these and just haven't gotten around to it yet, my apologies. Thanks again for the info.. ------- Liam Morley "light the deep, and bring silence to the world. light the world, and bring depth to the silence." |
