From lechner@cs.uml.edu Wed Oct 5 16:09:16 2005 Subject: Re: BDE and CHGEN - I'd like to put it at SourceForge. ANy ideas? To: sfrye@cs.uml.edu (Scott Frye) Cc: lechner@cs.uml.edu (Bob Lechner) RJLRef: $PH/COOL-GEN/sfryeChgenUses051005Addon.txt Re: BDE and CHGEN - I'd like to put it at SourceForge. ANy ideas? I just bought MBDoar: 'Practical SW Developmnt Environments', OReilly 2005 (290 pp paper, $40). It has broader scope than Fogel: OpenSource Dev. with CVS. (Coriolis 1999 $40 list). (Neither is about graphic IDEs.) R lechner > From sfrye@cs.uml.edu Wed Oct 5 15:35:16 2005 > To: Bob Lechner > Subject: Re: BDE and CHGEN - that's great Scott. Here's some info - > > Feel free to share my input. > > Are you planning on putting BDE up on SourceForge? or just leaving it on > the school system? > > (Notes below) > > -Scott Frye > > > On Wed, 5 Oct 2005, Bob Lechner wrote: > > > RJLRef: $PH/COOL-GEN/sfryeChgenUses051005.txt > > > > That's great, Scott. I'm happy that chgen is useful outside the class. > > I suspect gencpp/genjava would be better for you though. > > We may work on making them more compatible with each other > > and bde this semester. (4 of 5 students favor Java:-( > > I convinced them that genxml was merely a format converter > > (which could simplify style flexibility but is a lesser priority > > than robustness). > > > > Do you mind if I share your input with OOAD/SWE students and faculty? > > I hope GEN/LCP/BDE can become an open-source GPLicensed project > > When bde works again I'll add PKingston's WorkProc interface > > for DBDE ($PH/02s592?). > > > > I'd willing participate is such a project. I think the first major > challenge would be to select the implementation you would put forth as the > main branch and then generate a list of current and desired feartures for > the community to work on. The management of an open source project is > totally foriegn to me and I'm curious to see how it would work. > > > > $CASE holds an earlier chgen-based SQL project in which jomedina > > participated. Its goal was a Join operation using pfkeys. > > The test case was sparse matrix multiplication. > > See > > > > http://www.cs.uml.edu/~lechner/bde2sql/hrosario/user_manual.html > > > > Interactive acquisition of SQL queries instead of pre-compiled ones > > would be harder. Dynamic interpretation probably requires an > > interpreter of TT-->TA lists, like LCP inteerprets ST==>TR. > > > > Schema viewdefs could be extended with column (TA-subset) selection > > for display and report formatting. Then SQL parse trees could be > > converted to this VV--->VA<---TA association and interpreted > > dynamically. The interpreter like LCP could be a chgen-based > > application for query and report generation only (not DB update, > > which risks losing data irreversibly thru projection and selection, > > without more sophisication). > > > Instead of taking the track of representing the sql language, I took the > less ambitious route of just trying to represent 4 types of SQL > statements. 1) Create a new record given a key and assuming all other > values will be default. 2) Retrieving all fields of a single record from > a table that matches a key. 3) Update all fields of a single record > identified my the key. 4) delete a record that corresponds to a given key. > > I know this is limiting but I feel that the business logic of application > should be represented in the code, not the sql queries so these should be > sufficient for most uses. > > > Except for the importance of selection of TA_columns for projection > > (and graphic documentation?), I wonder what graphics adds to your > > application? E.g., for STDs and ERDs, both have hand-sketching > > plus text declaration workarounds that are faster than bde. It could be > > more productive to declare STDs and ERDs as text, then do the inverse > > map (sch2bde, SM2bde) to generate the documentation. > > True but I imagine it would be more desirable to create the graphs > directly in an environment instead of sketch->code as text->sch2bde. > The use of graphics limits the creative process to the tools available by > the graphic application whereas with a sketch a user is likely to develop > custome notation and then be faced with the problem of representing that > notation in the text declaration. > > > > > Bob Lechner > > > > > From sfrye@cs.uml.edu Wed Oct 5 10:18:00 2005 > > > Date: Wed, 5 Oct 2005 10:18:00 -0400 (EDT) > > > From: Scott Frye > > > To: Professor Lechner > > > Subject: BDE and CHGEN > > > Professor, > > > > > > I am interested in what steps the class is taking with BDE and chgen this > > > year. I take am taking Graphics II this semester and one of your > > > students (Jing) is in my class. He tells me that there has been some > > > discussions about making chgen and bde communicate with XML. > > > > > > In the last several months, I have become more interested in code > > > generation, specifically as it could apply to automating much of the > > > repetative coding that I do at my job. > > > > > > I wrote a graphic interface version of chgen (though not as extensive) > > > that allows me to select the fields and datatypes of a class, (with > > > buttons and listboxes), name a class and then generate the VB.NET code to > > > implements that class. The implementation is limited to Create, Retrieve, > > > Update, Delete (CRUD) functions. Last week I added a button that allows > > > me to generate the SQL scripts to create the table and stored procedures > > > in the MS-SQL database that corresponds to the class I generated in VB. > > > > > > Though the code is limited (and still a bit buggy), and not round trip, it > > > provides me with a tool that gives me code I can modify quickly to meet m > > > y needs. Earlier this week, I used the code to generate three classes for > > > a project that landed on my desk. The generated code was about 1000 lines > > > and would have taken me 2-3 days to write and debug by hand. > > > I generate it inside of 15 minutes and only spent an hour debugging it. > > > > > > Eventually, I would like to extend my solution to be more like BDE, but > > > this is a side project and my day job and school limit my time. > > > > > > Anyway, I thought you would be interested to know that I am applying the > > > techniques you taught me. > > > > > > -Scott Frye > > > > > >