Re: Re: OPEN REQUEST FOR RESURRECTING DOCUMENTATION
I'm interested in documentation, but know nothing about the internals (yet). Longer term I want to be able to use Dia functionality inother programs, esp. called from python, and that requires better understanding of the internals and the modularization. On Tue, 20 Mar 2001, Ben A. Hetland wrote: Date: Tue, 20 Mar 2001 20:51:01 +0100 To: [EMAIL PROTECTED], [EMAIL PROTECTED] From: "Ben A. Hetland" [EMAIL PROTECTED] Reply-To: [EMAIL PROTECTED] Subject: Re: OPEN REQUEST FOR RESURRECTING DOCUMENTATION Scott Harrison wrote: Fine with me. Nice set of pages you have there. So I guess the next big question is, WHO WANTS TO HELP DOCUMENT THINGS? If volunteers send me ([EMAIL PROTECTED]) their e-mail addresses, Well, I kinda already volunteered, so I'd like to contribute... although for the next month or so my first priority will be what is needed solely for my [payed] project. we can begin to set up a new effort at documentation. As soon (within a week) as there is some reasonable understanding as to how to proceed, we can begin communicating through dia-list as to what is going on. However, please note that the purpose of my original question was to locate documentation ideas that were already "there somewhere"... and to avoid the more tedious (and time-consuming) task of reverse-engineering. (Yes, I know that we programmer's tend to be sloppy about documenting for others what we make... gotta have some trade secrets... ;-) Hubert Figuiere wrote: [...snip...] Perhaps Doxygen (http://www.doxygen.org/) could be used as it allow to write code and doc at the same time ? That's an idea. But besides that, Doxygen might (?) have a problem identifying the classes buried in the Dia design, since after all it's C not C++ or Java... In my view, one of the problems with such automated tools to "resurrect and/or reverse engineer" existing code is that they're too accurate -- they include all the details and any patological interdependencies that might have grown into the source over the years, thereby obscuring the basic ideas structure. (It cannot distinguish between important and unimportant design features, ending up showing just a lot of "noise".) IMHO, merely converting code comments into HTML/RTF/whatever doesn't improve _that_ much over just reading the source files directly... It doesn't always reveal the structure. But that's just me, though... :-) Actually, I made a very tiny start using Dia itself (why not!?!!) and the UML elements, creating a class diagram based on header files and the README file. ...until I found that this was very time-consuming since I had not gained any familiarity with the source beforehand. -+-Ben-+- -- Harry George [EMAIL PROTECTED]
Re: Question about install dia in Solaries - need help!
I found that installing "gtkcanvas" helped get dia going on an SGI. The idea is that it is a subset of gnomecanvas, so you don't need anything more than glib, gtk, and gtkcanvas. On Fri, 09 Mar 2001, Ching-yu Hsueh wrote: Date: Fri, 09 Mar 2001 14:08:02 -0700 To: [EMAIL PROTECTED] From: Ching-yu Hsueh [EMAIL PROTECTED] Reply-To: [EMAIL PROTECTED] Sender: [EMAIL PROTECTED] Subject: Question about install dia in Solaries - need help! I do appreciate Is any body can help me to install dia in Sun Solaries? I tried to install glib-1.2.9 and gtk+-1.2.9 in my system first. Then tried dia-0.86 and had the error message that either gdk-pixbuf or imlib need to be installed. I don't know where to find it. (the ftp site had the file name without content) I tried to download and installed dia-0.2. When dia was executed, I had another error message "Couldn't find objectswhen looking for object-libs, exiting.". I had the message when I install dia, it said: "You must use the '-LLIBDIR' flag during linking. You will also need to do one of the following: --Add LIBDIR to the 'LD_LIBRARY_PATH' environmental variable during execution -- use the '-RLIBDIR' linker flag I don't know how to do it. Is it mean I have to use dia -RLIBDIR to execute the dia program? Any help from you shall be highly appreciated. I also wander if you can give me some idea where I can find the binary code to install dia in my Solaris (dia version and the file name). In this way I may have little trouble. Thanks! Ching-yu -- Harry George [EMAIL PROTECTED]
Re: Adding SQL support
I don't know anything about the internal code, but I was thinking of reading the XML dia file and using it to generate a) SQL, b) python code to read/write/query that DB, c) cgi's to provide a gui for the read/write. I did a bit of that for tcm, but ran out of time. (This time around I'll make no committments.) On Tue, 27 Feb 2001, Clifford Meece wrote: Date: Tue, 27 Feb 2001 22:12:25 -0600 To: [EMAIL PROTECTED] From: Clifford Meece [EMAIL PROTECTED] Reply-To: [EMAIL PROTECTED] Sender: [EMAIL PROTECTED] Subject: Adding SQL support Hi, I'm brand new to Dia, so please let me know if I'm treading on ground already covered. I would like to contribute to the dia project by adding data modeling support with an 'export to SQL' option. I may even build in the ability to go straight to the SQL server to create tables. Initially I would like to support MySQL, but eventually add support for other RDBS later. My initial thoughts were to take the existing code for UML support and modify it for an SQL library. Also, I can offer some website space for the Dia project if it is needed. Other areas that I would like to contribute are: 1. Automatic export from an SQL database of network infrastructure data to a Dia network Diagram. This will tie in to an existing project that I am working on. 2. Dia plugin for netscape. 3. A system so that Dia can be used for collaborative work. I'm interested in what the Dia community thinks of these ideas. Sincerely, Cliff Meece -- Harry George [EMAIL PROTECTED]
Re: dia MYSQL suggestion
UML is (necessarily) incomplete. But I think you can handle the data
already:
In type field, use the SQL type (or something which you can translate
to that)
In the value field, give the default value, plus indication of
NULL/NOT NULL and additional data as the "property string": From UML
User Guide, pg 128:
origin:Point=(0,0) {not null, frozen}
Becomes:
name: origin
type: Point
value: (0,0) {not null, frozen}
KEY is non-OO and thus non-UML, but could be done as a private
pseudomethod, "KEY", whose arguments are the list of attributes
needed to form the key.
On Thu, 8 Feb 2001, Andreas Prlic wrote:
Date: Thu, 8 Feb 2001 08:56:10 +0100
To: [EMAIL PROTECTED]
From: Andreas Prlic [EMAIL PROTECTED]
Reply-To: [EMAIL PROTECTED]
Subject: dia MYSQL suggestion
Hi!
I am using dia to draw physical entity relationship models of my
MYSQL
database. So I use the "UML class" object to create Tables and I
add
attributes to generate the columns of the Table.
At the moment the "attribute data" allows to specify attribute
name,
type and value. To properly describe the MYSQL database tables I
would need five "attribute data" fields, since in MYSQL every field
is
described by field name, type (int, char,etc) , NULL (if it can be
empty), Key , Default value and Extra info. Would it be possible to
create this kind of object?
Generalizing this one could imagine a table object, which the user
could define to be of N x M size, where M is the number of
attribute
data fields. And every data field could be named separately
greetings,
andreas
--
Andreas Prlic Center of Applied Molecular Engineering
University of Salzburg, Jakob-Haringerstr. 3
A - 5020 Salzburg, Austria
Tel: +43 - 662 - 8044 5798
http://www.came.sbg.ac.at/People/ap/home.html
--
Harry George
[EMAIL PROTECTED]
Re: OFFTOPIC: Looking for software modeling references
The first thing to know is that UML is just the latest in a long line of super-hyped silver bullets. It grew from the fact that 3 independent and competing salsemen/consultants found their business drying up (in the aftermath of the crash of CASE), and joined forces to make a unified product to sell. So start with the general process of software engineering, e.g., at: http://www.cs.queensu.ca/Software-Engineering/ Assuming you already know your way around scheduling, estimating, data modeling, testing, code reviews, etc., then we can talk UML. UML is mostly Rumbaugh's ideas, so reading his old book is a good start: "Object-oriented modeling and design", 1991. Booch, Rumbaugh, and Jacobson (the 3 salemen) wrote "The Unified Modeling Language User Guide: the ultimate tutorial to the UML", 1999. But that is almost worthless. Having read these, you will see there are diagrams for everything. Your job is to pick which ones you *need* for a given project -- the documentation plan. My rule of thumb is to use only verbal communication until/unless there is a specific need to do something more permanent. Then I use Dia, and put the diagram and the discussion of it (diagrams alone are not enough) on the web. Very seldom is there need for auto-tanslation from UML into actual code (e.g., Java, Python, SQL). Mostly, UML gets the humans talking the same general ideas, and then the code takes it the rest of the way to unification and completion. On 07 Feb 2001, [EMAIL PROTECTED] wrote: Date: 07 Feb 2001 15:29:30 -0800 To: [EMAIL PROTECTED] From: [EMAIL PROTECTED] Reply-To: [EMAIL PROTECTED] Sender: [EMAIL PROTECTED] Subject: OFFTOPIC: Looking for software modeling references I've been using dia for UML for some time now, but I am largely self-taught when it comes to modeling. Can anyone recommend some good references (dead trees, web, anything) for practical modeling with UML? I don't mean the syntax or symantics of UML, I mean the application of UML as a process for doing good modeling. Dave -- Harry George [EMAIL PROTECTED]
Fw: Re: Re: How to draw smart connectors in wiring diagrams?
(forwarded on behlf of Alan Gonzalez) On Wed, 31 Jan 2001, Alan Gonzalez wrote: Date: Wed, 31 Jan 2001 21:47:41 -0800 (PST) To: [EMAIL PROTECTED] From: Alan Gonzalez [EMAIL PROTECTED] Reply-To: [EMAIL PROTECTED] Subject: Re: Re: How to draw "smart" connectors in wiring diagrams? Ok, so this is similar to a prerouter i route. We are looking for close to shortest path, with a minimal amount of bends. Things like minimum spanning tree and shortest path can lead to a bit messy looking routes. The prerouter i wrote basically tries to keep things as straight as possible. So it will route the furthest items first, and then try to use this as a trunk and branch off this trunk, kinda of a ladder effect. One thing that makes this easier is when gridding is enabled so there are predefined tracks which the routes can take. In the case that gridding is not enabled, then i guess we can create an artificial smaller grid, which the user would not notice. Oh, BTW is Dia in C or C++. I haven't checked.. oh ok i see it's part of gnome so it's C. Hmm, well not especially my cup of tea, and also i was going to try to use the Graph Template Library which is part of Boost, but i guess that's out of the question. Most of the stuff i've done has been in C++ and am rusty with C. I'll look around at the sources and see if i can get into it. Alan --- Harry George [EMAIL PROTECTED] wrote: Clean, readable manhatten routing would be great, and reasonably eye-pleasing. True shortest path routing can look ugly. On Tue, 30 Jan 2001, Alan Gonzalez wrote: Date: Tue, 30 Jan 2001 22:41:42 -0800 (PST) To: [EMAIL PROTECTED] From: Alan Gonzalez [EMAIL PROTECTED] Reply-To: [EMAIL PROTECTED] Subject: Re: How to draw "smart" connectors in wiring diagrams? Hi, i have started playing around with Dia, and well haven't used Visio all that much. Anyways, i was a bit intrigued with the autorouting topic. When someone is saying autorouting, are we talking about placing/routing or just routing. As someone mentioned... the program "dot" goes through a lot of effort trying to make a "good/eye pleasing" placement for a logical grouping, the easier part of it is routing i believe. So if autorouting is simply having objects placed on the screen and you want connections automatically made, then that's relatively simple. I've written a VLSI router before (and am currently). In terms of someone's comment about magic, i wouldn't even go there since it's way too directed. For dia, i don't think we'll be needing congestion mapping, global routing then detailed routing and besides having all sorts of design rule violations. We can have some spacing rules, but you don't need all the types of things that magic provides. The only thing that dia complicates is the ability to have non-rectangular routes, so your dealing with a purely academic type of shortest path. I'm used to manhattan type routing where routes need to be 90 degree turns. This complicates things in terms of the intersection calculation routines, blockage detection. The graph can stay the same though. Anyways, i was just looking for a clarification. thanks. Alan Gonzalez __ Get personalized email addresses from Yahoo! Mail - only $35 a year! http://personal.mail.yahoo.com/ -- Harry George [EMAIL PROTECTED] __ Get personalized email addresses from Yahoo! Mail - only $35 a year! http://personal.mail.yahoo.com/ -- Harry George [EMAIL PROTECTED]
Re: Venn Diagrams with Dia ?
An alternative: Make the "universe rectangle". Make two circles, overlapping. Do dialogs...properties...draw background=no. You can stop there, label the regions, and be done. Or... Make a beziergon, with the control points at the intersecting lines of the two circles. Drag as best you can to match the curves. Change fill color. On Sat, 13 Jan 2001, Hilmar Strickfaden wrote: Hello Everybody I am currently discovering dia (newest linux-release) and I am very impressed by the possibilities one has with this powerful program. Due to the fact that I have to create some "Euler-Venn" diagrams (diagrams to display sets) for a student's script on mathematics I wonder if it is possible to create them with dia. I tried to draw a diagram in which two sets intersect themselves. Somehow it wasn't possible for me to create an object that represents the intersection set. Has anybody an Idea how I can use dia to create these Euler-Venn diagrams ? Try using a Beziergon and from the middle mouse menu choose 'Cusp control' for each endpoint. Then you can bend the endpoints to look like the intersection set. With judicious use of grid you can make various sets match up. Unfortunately, beziergons cannot create perfect circles, but you can approximate it. Two quick attempts can be seen at URL:http://shasta.cs.uiuc.edu/~lrclause/EulerVenn.dia -Lars -- Harry George [EMAIL PROTECTED]
Re: Middle mouse button on WINDOWS 98
Also, how about help using middle button on a 3 button mouse -- it thought it was a wheel. Setup: Toshiba 305 laptop. Win95. Mouse is a DEC 3 button (from a recycle bin). Just installed a bunch of cygwin stuff, but was running the new dia .exe download from the desktop icon, so I don't think that counts. Could be I just don't understand windows (thank gawd ;-) Old-Return-Path: [EMAIL PROTECTED] Date: Wed, 10 Jan 2001 13:06:18 +1300 From: Chris Goldie [EMAIL PROTECTED] X-Accept-Language: en Content-Type: text/plain; charset=us-ascii Resent-From: [EMAIL PROTECTED] Reply-To: [EMAIL PROTECTED] X-Mailing-List: [EMAIL PROTECTED] archive/latest/3155 X-Loop: [EMAIL PROTECTED] Precedence: list Resent-Sender: [EMAIL PROTECTED] X-UIDL: d5633f49ff90b26e1054dbbc99b2cc28 Anyone know how to access the middle mouse button when using a 2 button mouse with Dia Win32 port Any help would be appreciated Cheers Chris -- Harry George [EMAIL PROTECTED]
Re: Problem
A life line has two states: alive but asleep (dashed), and active (solid line, or skinny rectangle in dia). You re trying to communicate with the sleeping object. Try stretching out the rectangle and hooking lines to that. [You can do lots of messages that way. Also, you are using "message" for that, aren't you? ] On Fri, 22 Dec 2000, Jérôme Tamiotti wrote: Date: Fri, 22 Dec 2000 19:31:36 +0100 To: [EMAIL PROTECTED] From: Jérôme Tamiotti [EMAIL PROTECTED] Reply-To: [EMAIL PROTECTED] Sender: [EMAIL PROTECTED] Subject: Problem Hello, i'm very interested in using Dia for UML design, but i can't use it to create Sequences Diagrams, because it always crashes after having added the 2 life lines, and 1 or 2 messages... You'll see on the enclosed shot screen what happened Another problem is that it can't find some fonts, and opens many alert boxes.. Maybe these two problems are linked. Have you any idea ? Do you think it can come from my linux configuration ? I'm using Mandrake 7.2. Best regards, J.Tamiotti. -- Harry George [EMAIL PROTECTED]
Re: Re: [gnue-forms] Re: Embedded Scripting Language
I'm using python for many purposes and would really prefer that as the scripting language. Does anyone see show-stoppers either in license or in technology? Or is just a matter of someone steeping up to the task? On Sat, 25 Nov 2000, Ian D.Stewart wrote: Date: Sat, 25 Nov 2000 15:32:58 -0500 To: [EMAIL PROTECTED], [EMAIL PROTECTED] From: "Ian D.Stewart" [EMAIL PROTECTED] Reply-To: [EMAIL PROTECTED] Subject: Re: [gnue-forms] Re: Embedded Scripting Language On Fri, 24 Nov 2000, Derek A. Neighbors wrote: James Henstridge wrote: There is an incomplete python scripting plugin found in CVS. It is missing a number of features. It has a bit of trouble building at the moment because python only installs a static library, which libtool 1.3.x refuses to link into the .so for the plugin. Also, it is missing some of the properties interface code would be needed for it to be really useful. Hmm.. We have run into similar problems embedding python into our application server, this along with its license woes makes me think perhaps going to guile support would be a more wise decision. Any chance folks might be interested in JavaScript as a scripting language? There are GPL'ed implementations available in both C and Java. Regards, Ian D. Stewart -- "Rumors of my demise have been greatly exagerated" Samuel Clemens AKA Mark Twain -- Harry George [EMAIL PROTECTED]
Re: Re: ER diagrams; attributes
I'm not sure what you need. If you have selected the ER shape library, then make two Entities (rectangles), and then make 1 Relationship (diamond). Link via each entity to the relationship straight lines. Double-click the relationship diamond and edit its properties as needed. On 13 Nov 2000, [EMAIL PROTECTED] wrote: Date: 13 Nov 2000 17:00:19 -0800 To: [EMAIL PROTECTED] From: [EMAIL PROTECTED] Reply-To: [EMAIL PROTECTED] Subject: Re: ER diagrams; attributes I tried to use the ER set in DIA, but I'm confused about how to actually create relationships. Is there documentation or tutorial about this anywhere? Thanks, Dave -- Harry George [EMAIL PROTECTED]
Re: state machine
This is generally done in UML. The state is the circle with a black dot in the middle. Check the properties for additional options. The obvious transition line is the "message" line, but that doesn't allow for added segments (which it needs in general), so it doesn't do loop-back-on-self. So I use a spline, and add a segment so I can make a nice curve. Yes it is cumbersome, and I'd like to see a better solution. On Tue, 12 Sep 2000, Andreas Leitner wrote: Date: Tue, 12 Sep 2000 16:30:35 -0700 To: [EMAIL PROTECTED] From: Andreas Leitner [EMAIL PROTECTED] Reply-To: [EMAIL PROTECTED] Subject: state machine Hi, I am trying to draw a state machine with DIA. Are there any plug in available that suply me with some standard shades for it? Especially a transition from a state to itself is a little cumbersome to draw. many thanks in advance, Andreas -- Harry George [EMAIL PROTECTED]
Re: Troubles with Dia 0.86 + (GDK 1.2.8 and GTK+ 1.2.8)
I had trouble on sgi irix 6.2, also because I couldn't get gnome installed. But I found that installed just the gtk libs and then gtk-canvas did the trick. gtk-canvas is a repackaging of the graphics stuff (including gdk-pixbuf and libart) without the gnome overhead. On Fri, 25 Aug 2000, Juri Albert wrote: Date: Fri, 25 Aug 2000 14:54:57 +0200 To: [EMAIL PROTECTED] From: Juri Albert [EMAIL PROTECTED] Reply-To: [EMAIL PROTECTED] Subject: Troubles with Dia 0.86 + (GDK 1.2.8 and GTK+ 1.2.8) Hi all! I got troubles while compiling DIA on SPARC Ultra 5 Station with Solaris 8 running on it. All required libraries except gdk-pixbuf, because it needs the Gnome libraries to be installed But ./configure ran fine. After i start make and gcc compiles about 30 - 40 minutes Here the error message I finally see and can't get around it: /usr/local/include/gdk/gdktypes.h:1045: storage size of `area' isn't known /usr/local/include/gtk/gtkwidget.h:213: storage size of `requisition' isn't known /usr/local/include/gtk/gtkclist.h:159: storage size of `internal_allocation' isn't known /usr/local/include/gtk/gtkclist.h:170: storage size of `column_title_area' isn't known /usr/local/include/gtk/gtkfontsel.h:162: storage size of `filters' isn't known make[2]: *** [connectionpoint_ops.o] Error 1 make[2]: Leaving directory `/home/III/stud/albert/dia-0.86/app' make[1]: *** [all-recursive] Error 1 make[1]: Leaving directory `/home/III/stud/albert/dia-0.86' make: *** [all-recursive-am] Error 2 Has anyone an idea, how to solve this problem ? The full listing of my make procedure I included as attachment to my mail And another question: Has anybody already compiled dia under Solaris 7/8 on SPARC or even X86 Plattform ? Thanks. With best regards, Juri Albert -- Harry George [EMAIL PROTECTED]
Dia tutorial
I've been using Dia, but haven't found a tutorial. So I made one. See: http://www.seanet.com/~hgg9140/comp/index.html Is there something better than this? (e.g., hidden in the Dia Documentaition project?) If not, you can have this one. BTW: Dia is great. Thanks you for saving me from Visio :-). -- Harry George [EMAIL PROTECTED]
