Re: DFD plugin
On Fri, 18 May 2001, Mark Cotter wrote: I am new to using DIA having just downloaded it about 2 hours ago and I am already amazed by what it can do. I have already put together a UML diagram for a project that I am doing that I just had written out on paper - time to throw away the notepad I think. We're happy you like it. Please let us know of any bugs/ideas you come upon at bugzille.gnome.org. One thing that I would have liked to have seen would have been a DFD plugin. Does anyone know if there is such a plugin? Not that I've seen, and most plugins go onto the list. You shall be more than welcome to make one -- the FAQ has some information on how to make simple shapes: URL:http://www.lysator.liu.se/~alla/dia/faq.html -Lars -- Lars Clausen (http://shasta.cs.uiuc.edu/~lrclause) | Hårdgrim of Numenor I do not agree with a word that you say, but I| Retainer of Sir Kegg will defend to the death your right to say it.| of Westfield --Evelyn Beatrice Hall paraphrasing Voltaire | Chaos Berserker of Khorne
Re: (Was: How to create shapes on the fly?)
On Fri, 18 May 2001, Cyrille Chepelov wrote: What could be desirable would be a way to edit the properties of a member of a group -- but I don't know how to handle editing the properties of a member of a group, itself a member of a larger group, with a way to do this intuitive enough so it's useful to users. XFig has the feature when you can 'open' groups. Then you can only see that group and edit the items singly, then later close it. We might want something somewhat similar, but maybe keeping the rest shown, marking open groups with dotted lines. I've also heard requests for remembering groups when ungrouping. -Lars -- Lars Clausen (http://shasta.cs.uiuc.edu/~lrclause) | Hårdgrim of Numenor I do not agree with a word that you say, but I| Retainer of Sir Kegg will defend to the death your right to say it.| of Westfield --Evelyn Beatrice Hall paraphrasing Voltaire | Chaos Berserker of Khorne
Re: triaging bugs ?
On Wed, 16 May 2001, Hubert Figuiere wrote: I'd like to help sort out bugs for Dia in bugzilla.gnome.org. There are several that seems to be fixed and some that I'd wish to comment. The only problem is that I'm only allowed to change bugs I did submit. What are the steps to request an account that have enough privileges ? In fact, most are fixed, I went over them recently. Check my mails from two weeks or so ago. -Lars -- Lars Clausen (http://shasta.cs.uiuc.edu/~lrclause) | Hårdgrim of Numenor I do not agree with a word that you say, but I| Retainer of Sir Kegg will defend to the death your right to say it.| of Westfield --Evelyn Beatrice Hall paraphrasing Voltaire | Chaos Berserker of Khorne
Re: How to create shapes on the fly?
On Wed, 16 May 2001, Matti Picus wrote: I saw that xfig uses groups and a nice format for its parser, I will try to rewrite the DXF importer along those lines. Thank you! Ironicallz, the xfig importer was written with inspiration from the DXF importer. At 07:42 AM 5.16.2001 +0200, Steffen Macke wrote: Hi, scaling, and rotation. The current impelmentation of the dxf-import.c ignores all blocks. From my two hours of looking a the source tarball from Dia 0.88, it seems that the parallel structure in Dia would be shapes. Wouldn't groups be enough? The dxf-import could then look up the BLOCK in the DXF and draw it as a group everywhere it is inserted. Isn't this wasteful in terms of memory, and the possiblity to modify once and have the change reflected in many insertions of the same group? Yes, it would be inferior in that way. Shapes don't get stored along with the dia file - they would be lost after saving the dxf as a dia file. Too bad. What would it involve to allow shapes to be stored in a dia file? Well, export to shape files does that already, so look that up. Also, the plugin menu of the toolbox allows loading modules, so you could have a 'dynamic' module that is reloaded every time a new shape is created. It's all in there. In fact, it would be useful overall to, say, have a 'Create Shape' menupoint (creating shapes dynamically from objects/groups). -Lars -- Lars Clausen (http://shasta.cs.uiuc.edu/~lrclause) | Hårdgrim of Numenor I do not agree with a word that you say, but I| Retainer of Sir Kegg will defend to the death your right to say it.| of Westfield --Evelyn Beatrice Hall paraphrasing Voltaire | Chaos Berserker of Khorne
Re: Dia and fonts
On Fri, 11 May 2001, Cyrille Chepelov wrote: Le Fri, May 11, 2001, à 07:47:28PM +0900, Richard Sharpe a écrit: However, Dia shows me a very restricted list of fonts, it seems, when I insert text. How can I get Dia to show me them all, like Verdana, etc. This has been intentionally restricted. The short reason why, is that it's quite difficult to match the fonts available under X and the fonts available for printing. Portability of documents could be cited as another reason, since it would also be difficult to retrieve the font file, embed it into a dia file, and deliver the result on another machine while loading the diagram, to display or print it. This will probably last, until the usual chaos about fonts under X/PS-using desktops gets cleared, preferably with a solution which doesn't displease too many people... (once *that* is done, there will necessarily be some modifications to do to dia to take advantage of this, but compared with the unified font system itself, it'll probably be a snap to do). I disagree on this; I think we should get some better font support soon, even if it means that some documents will use replacement fonts when moved between machines. PostScript output can just have the font embedded. The problem of different font installations on different machines will not get fixed until there is some kind of font-server available for all that allows the document to specify and download fonts, and I don't see that happening any time soon. So if we can deal with the PostScript problems, we should be able to use whatever fonts are installed. We could even mark the fonts that are not PostScript compatible until such time as we can embed fonts into PostScript. -Lars -- Lars Clausen (http://shasta.cs.uiuc.edu/~lrclause) | Hårdgrim of Numenor I do not agree with a word that you say, but I| Retainer of Sir Kegg will defend to the death your right to say it.| of Westfield --Evelyn Beatrice Hall paraphrasing Voltaire | Chaos Berserker of Khorne
Re: Dia and fonts
On Fri, 11 May 2001, Cyrille Chepelov wrote: Le Fri, May 11, 2001, à 10:33:27AM -0500, Lars Clausen a écrit: I disagree on this; I think we should get some better font support soon, even if it means that some documents will use replacement fonts when moved between machines. PostScript output can just have the font embedded. How do you retrieve the PS fonts' outlines ? (assuming we'll let TTF aside for the moment) ? How do you even ask the font serving service (I'm deliberately avoiding the term font server, for obvious reasons) for where the file are stored, or if that service is not located on the same machine as dia or as the display, how do you download these files ? [...] I did a little digging, and found the FreeType library URL:www.freetype.org. It allows manipulating various vector font formats, including Type 1 and TrueType. It's very portable and most Linux distributions have it. I haven't found anything yet about rendering to PostScript, but I found a TrueType to Type 1 converter library URL:http://ttf2pt1.sourceforge.net that could give us the code if we need. Given that this would add another library dependence, I think it should be optional, but I'm willing to put some time into getting it up. Not for this release, obviously, but definitely before 1.0. I haven't looked into how well it works, but it sounds better than waiting for Pango and the like. I've seen installations (BSD's in particular) that didn't have all the standard PostScript fonts for X, so only a few font selections were available. Is that better than just using the X fonts and risk worse printing? -Lars -- Lars Clausen (http://shasta.cs.uiuc.edu/~lrclause) | Hårdgrim of Numenor I do not agree with a word that you say, but I| Retainer of Sir Kegg will defend to the death your right to say it.| of Westfield --Evelyn Beatrice Hall paraphrasing Voltaire | Chaos Berserker of Khorne
Re: Dia and fonts
On 11 May 2001, Lars Clausen wrote: On Fri, 11 May 2001, Cyrille Chepelov wrote: Le Fri, May 11, 2001, à 10:33:27AM -0500, Lars Clausen a écrit: I disagree on this; I think we should get some better font support soon, even if it means that some documents will use replacement fonts when moved between machines. PostScript output can just have the font embedded. How do you retrieve the PS fonts' outlines ? (assuming we'll let TTF aside for the moment) ? How do you even ask the font serving service (I'm deliberately avoiding the term font server, for obvious reasons) for where the file are stored, or if that service is not located on the same machine as dia or as the display, how do you download these files ? [...] I did a little digging, and found the FreeType library URL:www.freetype.org. It allows manipulating various vector font formats, including Type 1 and TrueType. Looking through the mail archives for PostScript output, I notice that the FreeType library does string rotation. We want that, too. -Lars -- Lars Clausen (http://shasta.cs.uiuc.edu/~lrclause) | Hårdgrim of Numenor I do not agree with a word that you say, but I| Retainer of Sir Kegg will defend to the death your right to say it.| of Westfield --Evelyn Beatrice Hall paraphrasing Voltaire | Chaos Berserker of Khorne
Re: Dia and fonts
On Fri, 11 May 2001, Cyrille Chepelov wrote: Le Fri, May 11, 2001, à 02:56:29PM -0500, Lars Clausen a écrit: Given that this would add another library dependence, I think it should be optional, but I'm willing to put some time into getting it up. Not for this release, obviously, but definitely before 1.0. You're not actually advocating rasterising the fonts ourselves, are you ? However, if you really are, then maybe this would give rotated text (kiddingI've already done this, bwahahahaha/kidding). No, the FreeType library renders for us. Including rotating^H^H^Hed text. Simple problem: how do you know what the font path is (callparse xset q, OK), and how to access the non-local fonts (ban XFS-accessed fonts, acceptable if we ban X terminals, OK). I don't, FreeType does. (another, a bit dirty but probably effective *and* easy solution to render Type1 fonts would be to simply download them into the Display PS renderer... we keep some of X's elegance, but we still add a bunch dependencies (DPS, and a DPS client. We'd get bitten a bit by the latter, since DPS-X is not yet easily packageable (because of Freeness issues with the AFPL half), and the former DGS as done by the Gyve project has been unpackaged (dropped) in Debian. Sucks but true). When I tried it out, it didn't seem ready at all. Lots of bugs and crashes. I haven't looked into how well it works, but it sounds better than waiting for Pango and the like. We *will* have to use Pango when we can use it, anyway. I don't know for you, but I really don't want to handle i18n and multiple scripts myself. The Pango team has done it, already (well, sort of, otherwise we'd already just linked it up). How much difference will it be to use Pango? I would like to have all text handling encapsulated (no calls to gtk_text_width etc), so we can easily pick. And note that I'm suggesting this as an optional library. I've seen installations (BSD's in particular) that didn't have all the standard PostScript fonts for X, so only a few font selections were available. Is that better than just using the X fonts and risk worse printing? What worse printing ? I'd think raster fonts are totally out of the discussion, aren't they ? Well, you're discussing them above, aren't you:) For one thing, a high DPI rastering is preferable to not having the font at all. Second, from what I see in the FreeType mailing list, it handles conversion to Type, so we wouldn't need to raster for PostScript. Well, I'm about to head off to Europe for three weeks. I'll be in intermittent email contact. I have no problems with the current tarball, though I hope all of you will bang at it till it compiles jcleanly under all configuration combos. -Lars -- Lars Clausen (http://shasta.cs.uiuc.edu/~lrclause) | Hårdgrim of Numenor I do not agree with a word that you say, but I| Retainer of Sir Kegg will defend to the death your right to say it.| of Westfield --Evelyn Beatrice Hall paraphrasing Voltaire | Chaos Berserker of Khorne
Groups and depths
Here's a problem with groups I just noticed: Grouping objects destroys their depths. When you create a group, it becomes a new object on top of the other objects, and the depth of the objects is ignored. Try this: Create three partially overlapping rectangles. Select the two lowest. Group them. They jump to the front. The objects in the group retain their relative depth, which is good. But the group object should not really have a depth of its own, rather the individual objects should be at their original depths. (And the Group/Ungroup operations should be idempotent inverses.) That, however, is tricky to do with the current implicit ordering. Is this something we should work on? Have anyone been bothered by this? I only noticed now that I look at groups for XFig. -Lars -- Lars Clausen (http://shasta.cs.uiuc.edu/~lrclause) | Hårdgrim of Numenor I do not agree with a word that you say, but I| Retainer of Sir Kegg will defend to the death your right to say it.| of Westfield --Evelyn Beatrice Hall paraphrasing Voltaire | Chaos Berserker of Khorne
Xfig import doesn't crash now (+intl)
The good news is that Xfig import no longer crashes on any of my samples. It should probably be marked as incomplete, though, as it doesn't do polylines and beziers and such. The bad news is that I noticedabout 70 instances of message_error(), message_warning() or message_notice() being called without the string being in _(). -Lars -- Lars Clausen (http://shasta.cs.uiuc.edu/~lrclause) | Hårdgrim of Numenor I do not agree with a word that you say, but I| Retainer of Sir Kegg will defend to the death your right to say it.| of Westfield --Evelyn Beatrice Hall paraphrasing Voltaire | Chaos Berserker of Khorne
Re: Define closed object with different types of line segments ...
On Fri, 11 May 2001, Richard Sharpe wrote: Hi, I would like to be able to define an object made up of different types of line segments (beziers, straight, curved, etc) ... Then it would be nice to have that object available as something I could add anywhere, rescale it, etc. I don't want to have to write a progam for it. I would also like to fill the object ... Use the Beziergon, and use the middle menu to make the midpoints Cusp controls. That will give you the straight lines. It won't quite give you curved lines, but you can get pretty close. There is no way to define a filled multi-line object the way you describe, as fill is implemented with filled boxes, ellipses, and beziers rather than a fill-in. -Lars -- Lars Clausen (http://shasta.cs.uiuc.edu/~lrclause) | Hårdgrim of Numenor I do not agree with a word that you say, but I| Retainer of Sir Kegg will defend to the death your right to say it.| of Westfield --Evelyn Beatrice Hall paraphrasing Voltaire | Chaos Berserker of Khorne
Re: Groups and depths
On Fri, 11 May 2001, James Henstridge wrote: On 10 May 2001, Lars Clausen wrote: Here's a problem with groups I just noticed: Grouping objects destroys their depths. When you create a group, it becomes a new object on top of the other objects, and the depth of the objects is ignored. Try this: Create three partially overlapping rectangles. Select the two lowest. Group them. They jump to the front. The objects in the group retain their relative depth, which is good. But the group object should not really have a depth of its own, rather the individual objects should be at their original depths. (And the Group/Ungroup operations should be idempotent inverses.) That, however, is tricky to do with the current implicit ordering. Is this something we should work on? Have anyone been bothered by this? I only noticed now that I look at groups for XFig. I always considered groups as being a single object. It sounds a bit weird for another object to be `in between' a group. Maybe creating a group shouldn't cause the objects to raise to the top of the stack, but grouped objects have always acted as one position in the depth stack in all the drawing software I have used. It depends on what a group is. It makes sense to say a group is a bunch of objects you manipulate together, in which case grouping shouldn't change their depth. But I know how hard it would be to change that in Dia, and I got a reasonable fix for XFig import (it puts it at the highest depth involved), so I don't think we should worry. -Lars -- Lars Clausen (http://shasta.cs.uiuc.edu/~lrclause) | Hårdgrim of Numenor I do not agree with a word that you say, but I| Retainer of Sir Kegg will defend to the death your right to say it.| of Westfield --Evelyn Beatrice Hall paraphrasing Voltaire | Chaos Berserker of Khorne
Re: Arrows as part of lines, argument for
On Wed, 9 May 2001, [EMAIL PROTECTED] wrote: hi, wouldn't it be nice to have fallbacks in RenderOps, lika a DrawArrowFunc, which would only be called, when set. Otherwise lines could be used. the same way you could think of other fallbacks, for recangles, lines, if there would be a DrawPointFunc, etc. this way you could implement a new renderer with only implementing a very basic set of functions, eg a line. Interesting idea -- have drawRectangle, drawPolyline etc fall back to drawLine could be useful. But for arrowheads, it would require a very different approach, since the arrow information is still separate from the line information. i think this way it would be very easy, for example to write graphic format renderers. That would help get a trivial renderer up. -Lars -- Lars Clausen (http://shasta.cs.uiuc.edu/~lrclause) | Hårdgrim of Numenor I do not agree with a word that you say, but I| Retainer of Sir Kegg will defend to the death your right to say it.| of Westfield --Evelyn Beatrice Hall paraphrasing Voltaire | Chaos Berserker of Khorne
Re: dia-0.87.tar.gz available
On Wed, 9 May 2001, James Henstridge wrote: After getting all the make distcheck bugs out of the way, I have put together the new dia tarball. It will be available at: ftp://ftp.gnome.org/pub/GNOME/stable/sources/dia/dia-0.87.tar.{gz,bz2} (once ftp.gnome.org syncs with master.gnome.org). I haven't announced this very widely yet, but if no one finds anything serious, I will put out announcements tomorrow. Ok, let the good bugs roll: When compiling with unicode support, it lacks lib/ps-utf8.[ch]. I know it's experimental, but it should at least be there:) plugins/python is missing python-startup.py. make install complains when making the docbook: jade:../dia-manual.sgml:1:66:W: cannot generate system identifier for public text -//GNOME//DTD DocBook PNG Variant V1.1//EN jade:../dia-manual.sgml:16:0:E: reference to entity BOOK for which no system identifier could be generated jade:../dia-manual.sgml:1:0: entity was defined here jade:../dia-manual.sgml:16:0:E: DTD did not contain element declaration for document type name jade:../dia-manual.sgml:18:9:E: there is no attribute ID jade:../dia-manual.sgml:18:16:E: element BOOK undefined ... The xfig import plug-in is somewhat broken: It will load most things, but can cause a crash when closing the diagram. I'll go a-hunting. Other than that, it looks fine. Now we can start looking at what should go into the next release. -Lars -- Lars Clausen (http://shasta.cs.uiuc.edu/~lrclause) | Hårdgrim of Numenor I do not agree with a word that you say, but I| Retainer of Sir Kegg will defend to the death your right to say it.| of Westfield --Evelyn Beatrice Hall paraphrasing Voltaire | Chaos Berserker of Khorne
How to create a group?
I've narrowed the bug in xfig-import.c down to the handling of groups. Everything else behaves nicely (well, doesn't crash at least), but whenever there is a group involved, it crashes when the window gets closed. This is the function I use to create a group. The items are all already added to the layer, and work fine outside of groups. Am I missing something crucial about groups? static Object * create_standard_group(GList *items, DiagramData *dia) { Object *new_obj; new_obj = group_create((GList*)items); layer_add_object(dia-active_layer, new_obj); return new_obj; } -Lars -- Lars Clausen (http://shasta.cs.uiuc.edu/~lrclause) | Hårdgrim of Numenor I do not agree with a word that you say, but I| Retainer of Sir Kegg will defend to the death your right to say it.| of Westfield --Evelyn Beatrice Hall paraphrasing Voltaire | Chaos Berserker of Khorne
Re: How to create a group?
On Thu, 10 May 2001, James Henstridge wrote: On 9 May 2001, Lars Clausen wrote: I've narrowed the bug in xfig-import.c down to the handling of groups. Everything else behaves nicely (well, doesn't crash at least), but whenever there is a group involved, it crashes when the window gets closed. This is the function I use to create a group. The items are all already added to the layer, and work fine outside of groups. Am I missing something crucial about groups? Try not adding them to the layer? That's part of the answer. Looking at the load_save code, I think not adding the component parts to the layer is what I want. Now I'll just have to make that work with the xfig depth model. -Lars -- Lars Clausen (http://shasta.cs.uiuc.edu/~lrclause) | Hårdgrim of Numenor I do not agree with a word that you say, but I| Retainer of Sir Kegg will defend to the death your right to say it.| of Westfield --Evelyn Beatrice Hall paraphrasing Voltaire | Chaos Berserker of Khorne
Re: plug-in xfig for dia
On Tue, 08 May 2001, Jamal Malki wrote: Bonjour, I write you to ask wether there exists a plug-in xfig for exporting dia drawing into the xfig format file. Not yet, but it wouldn't be difficult to make. I have some xfig import already in place (though buggy). I doubt I can get export working before I take off for Denmark at the end of the week, though. -Lars -- Lars Clausen (http://shasta.cs.uiuc.edu/~lrclause) | Hårdgrim of Numenor I do not agree with a word that you say, but I| Retainer of Sir Kegg will defend to the death your right to say it.| of Westfield --Evelyn Beatrice Hall paraphrasing Voltaire | Chaos Berserker of Khorne
Arrows as part of lines, argument for
Here's an argument for having the arrowhead be part of the line rather than slapped on top: When exporting to a format that has lines with arrows, you lose the arrowhead information, and the arrow becomes a set of lines that just happen to be at the end of the line. -Lars -- Lars Clausen (http://shasta.cs.uiuc.edu/~lrclause) | Hårdgrim of Numenor I do not agree with a word that you say, but I| Retainer of Sir Kegg will defend to the death your right to say it.| of Westfield --Evelyn Beatrice Hall paraphrasing Voltaire | Chaos Berserker of Khorne
Re: plug-in xfig for dia
On 08 May 2001, Lars Clausen wrote: On Tue, 08 May 2001, Jamal Malki wrote: Bonjour, I write you to ask wether there exists a plug-in xfig for exporting dia drawing into the xfig format file. Not yet, but it wouldn't be difficult to make. I have some xfig import already in place (though buggy). I doubt I can get export working before I take off for Denmark at the end of the week, though. So I lied. It's now in CVS, except for splines (which seem to be less powerful in XFig than in Dia, see URL:http://epb1.lbl.gov/xfig/drawing.html#spline. XFig import also made it into the menus now, but it's somewhat broken, so unless somebody else wants to look at it, it should probably no be in the release. -Lars -- Lars Clausen (http://shasta.cs.uiuc.edu/~lrclause) | Hårdgrim of Numenor I do not agree with a word that you say, but I| Retainer of Sir Kegg will defend to the death your right to say it.| of Westfield --Evelyn Beatrice Hall paraphrasing Voltaire | Chaos Berserker of Khorne
Re: Dia SQL export plug-in available
On Mon, 7 May 2001, Andrew S. wrote: There is a Dia SQL export plug-in available at: ftp://ftp.tuxtopia.com/pub/dia/plug-ins/sql.tar.gz ftp://az.water.usgs.gov/pub/ashalper/src/dia/plug-ins/sql.tar.gz It currently only supports UML class diagrams, and can only generate plain SQL CREATE TABLE statements. The next version will support the { PRIMARY KEY | UNIQUE } syntax. Both of these FTP servers are probably quite flakey. If someone would like to add it to the CVS tree, I would be happy to maintain it there. How does this compare to the Dia2SQL project on SourceForge? -Lars -- Lars Clausen (http://shasta.cs.uiuc.edu/~lrclause) | Hårdgrim of Numenor I do not agree with a word that you say, but I| Retainer of Sir Kegg will defend to the death your right to say it.| of Westfield --Evelyn Beatrice Hall paraphrasing Voltaire | Chaos Berserker of Khorne
Re: Is this a Known Bug ????
On Thu, 03 May 2001, Ben A. Hetland wrote: Cyril ZEKSER wrote: I've made an UML diagram, with classes and hierarchy ! When I open it, the connector aren't connected correctly (as I saved them) , as shown in the sample below : [Image] Hmmm... do you mean that the lines aren't connected at the blue cross location, as they probably should? Otherwise they seem nicely lined-up. If so, I think I have observed a similar erraneous behavior when working with a diagram on multiple computers having different screen resolution (and also Win32-Linux). In my case however, the lines' end points just end up far away from the element it's connected to, and they look like they're not connected at all (but they are!). This is only initially after opening the file -- as soon as I move the element around, the line pops into place again. Annoying... I could imagine this being caused by fonts being of different size on different platforms. Does this also happen when no text is involved? -Lars -- Lars Clausen (http://shasta.cs.uiuc.edu/~lrclause) | Hårdgrim of Numenor I do not agree with a word that you say, but I| Retainer of Sir Kegg will defend to the death your right to say it.| of Westfield --Evelyn Beatrice Hall paraphrasing Voltaire | Chaos Berserker of Khorne
Re: dia doesn´t load my files
On Thu, 03 May 2001, Florian Berger wrote: Hi guys, I´m using dia and it really is a great thing to work with. Even ground-plans are no problem. ;-) What´s wrong: When I save a file (uncompressed xml), quit dia, start again and try to load the file again, dia generates about 50 or 80 windows saying Error parsing point. It finally shows the grid, but no element is drawn, so in the end I´m not able to open a file again. I´s using SuSE Linux 7.0 Kernel 2.2.16 with GNOME 1.2.4 and DIA 0.86. This is a known problem in Dia 0.86 that is fixed in the current CVS version and thusly in the upcoming release. It is caused by the decimal point of floating-point numbers being output in your locale (German, yesno?) in the XML file. Either compile the newest CVS version (which has a bunch of other improvements), or wait a bit for the next release (which should be coming up soon). -Lars -- Lars Clausen (http://shasta.cs.uiuc.edu/~lrclause) | Hårdgrim of Numenor I do not agree with a word that you say, but I| Retainer of Sir Kegg will defend to the death your right to say it.| of Westfield --Evelyn Beatrice Hall paraphrasing Voltaire | Chaos Berserker of Khorne
Re: Textbox filled by default with some text
On Thu, 3 May 2001, Rutger Nijlunsing wrote: Hi all, I want to add several shapes with text already filled in in the shape, preferably non-edittable. I see several solutions to this: * Add a field 'text' to the textbox tag: textbox x1=left y1=top x2=right y2=bottom text=Sample text/ ...where Sample text is the default to initialise the text box with, when the shape is instantiated. * Add to the SVG parser a way to parse SVG's text fields like svgtextSample text/text/svg. * Convert the text to vector graphics, convert it to svg, and add those polylines to the shape file. I prefer the first one, since it sounds quite simple to do. Yes. That would of course make an editable text field. Maybe a read-only flag as well? Anyone knows how and where in the code to implement it? The code you are looking for is in objects/custom/shape_info.c, line 893+. -Lars -- Lars Clausen (http://shasta.cs.uiuc.edu/~lrclause) | Hårdgrim of Numenor I do not agree with a word that you say, but I| Retainer of Sir Kegg will defend to the death your right to say it.| of Westfield --Evelyn Beatrice Hall paraphrasing Voltaire | Chaos Berserker of Khorne
Re: Is this a Known Bug ????
On Thu, 03 May 2001, Ben A. Hetland wrote: Lars Clausen wrote: On Thu, 03 May 2001, Ben A. Hetland wrote: [...] If so, I think I have observed a similar erraneous behavior when working with a diagram on multiple computers having different screen resolution (and also Win32-Linux). In my case however, the lines' end points just end up far away from the element it's connected to, and they look like they're not connected at all (but they are!). This is only initially after opening the file -- as soon as I move the element around, the line pops into place again. Annoying... I could imagine this being caused by fonts being of different size on different platforms. Does this also happen when no text is involved? I have mainly used the UML elements when I discovered this, but I'm not sure if it occurred on every possible combination of migrating the files. (At the moment I'm a bit stuck with a half-empty brand new machine... so I have a lot to reinstall.) I.e., I observed it especially on the right edges of UML Class objects, but they definitely do involve fonts, as you point out. They also have the property that they determine their own size (depending on font?). Hmmm... they seem to calculate everything correctly. Could you send a small example diagram with this behaviour (preferably just one object that has wrong size)? If you send me one from the Windows machine, I should be able to see it, too. -Lars -- Lars Clausen (http://shasta.cs.uiuc.edu/~lrclause) | Hårdgrim of Numenor I do not agree with a word that you say, but I| Retainer of Sir Kegg will defend to the death your right to say it.| of Westfield --Evelyn Beatrice Hall paraphrasing Voltaire | Chaos Berserker of Khorne
Re: New version soon?
On Tue, 1 May 2001, Aaron Trevena wrote: On Tue, 1 May 2001, Hans Breuer wrote: - work on all major platforms, (Does this include Gnome 1.4; anyone using Gnome 1.4 out there ?) - make available all the bugfixes since the last official version (Dia 0.86 has had some serious problems with some menu actions, see: Objects/Align/Equal Distance thread. Are there any *users* out there ?) - contain less bugs/more features than the previous release - silence complaining about bugs, which got fixed month ago - be localized properly ? I think one of the biggest features would be increasing extensibility or providing more and easier to find documentation on extending dia, be it plugins or tools that work along side Dia. Yes, better organization of sheets and stuff would be neat-o. but thats just me because I never want to use Visio again, actually I never want to use any diagram prgram again I want it all to work magically thru perl scripts and xml and mind reading, but one step at a time :) I almost got my coworker here to write the mind-reading plugin:) as for releasing early and often - thats what CVS is for amongst the other good points hans makes. No, CVS is for the developers. Having to point non-developers at the CVS version is IMHO a sign of lack of frequent enough releases. Most importantly, distributions and the like should never pick from CVS (see what happened to RedHat and gcc), but should be able to include a recent version. One of the things that makes a software project more dynamic is having making it more accesable through plugins and api's and documentation that mean people can pick it up and start hacking. I'd love to hack dia but my C isn't too hot, hence me doing automagic with Dia's XML using perl, having an API or a way to do perl plugins would make a huge difference. Let me put it this way: What do you want to be able to do from perl? Just create more objects, or more than that? finally thanks to the Dia developers and maintainers for doing things the right way - the XML file format is really useful - I am using it to create UML class diagrams in perl and now I'm parsing the xml into text/html documentation as well. I know of a bunch of other projects that taking the xml and doing cool stuff with it like dia2sql. Thanks for that go to Alex -- he did a very cool design right from the start, including the split between app, lib and objects. Anyone got any thoughts on Graphviz - it does automagic layout, don't know how much use or interest it would be to the Dia developers? I am planning to make my code work with both Dia and Graphviz. I believe the Graphviz stuff is under a rather more restrictive license. -Lars -- Lars Clausen (http://shasta.cs.uiuc.edu/~lrclause) | Hårdgrim of Numenor I do not agree with a word that you say, but I| Retainer of Sir Kegg will defend to the death your right to say it.| of Westfield --Evelyn Beatrice Hall paraphrasing Voltaire | Chaos Berserker of Khorne
Re: New version soon?
On Tue, 01 May 2001, Hans Breuer wrote: At 08:56 30.04.01 -0500, Lars Clausen wrote: On Mon, 30 Apr 2001, Claus Sørensen wrote: What is the roadmap for the next version? Let's see if all the developers/users can agree on this ? Dia 0.87 should: - work on all major platforms, (Does this include Gnome 1.4; anyone using Gnome 1.4 out there ?) - make available all the bugfixes since the last official version (Dia 0.86 has had some serious problems with some menu actions, see: Objects/Align/Equal Distance thread. Are there any *users* out there ?) Yes, I use it for a variety of stuff (from DTP-like setup through scenarios and flowcharts for games to actual diagrams for articles). - contain less bugs/more features than the previous release I believe we do. We haven't removed any features that I know of, and I put in the feature that it shows the object name in the properties dialog:) - silence complaining about bugs, which got fixed month ago - be localized properly ? That's a toughie -- translating all the sheets require knowledge not only of the language, but of the specific terms. I've been bugging various LUGs to get translators, but no luck yet. Is there a feature freeze? Interesting question. Lars, what do you think about this one ? Yes, I've had the feeling we've had feature freeze for a while. I even think it was mentioned on the list once. Couldn't you release more often so more of us can benefit for the work you have been doing since last release. IHMO the key question here is: Is the benefit on both sides ? Doing an official release does require some extra work which almost never falls into the fun category. Part of this work can be done by the users, who are willing to build from cvs. Doing an official release should also relieve us from a) having to point users at CVS for bugs that got fixed months ago, and thus b) having to help non-developers with getting the CVS version to compile. There will be a new version soon, even if we have to go to Australia ourselves and hunt down James:) -Lars -- Lars Clausen (http://shasta.cs.uiuc.edu/~lrclause) | Hårdgrim of Numenor I do not agree with a word that you say, but I| Retainer of Sir Kegg will defend to the death your right to say it.| of Westfield --Evelyn Beatrice Hall paraphrasing Voltaire | Chaos Berserker of Khorne
Re: New version soon?
On Tue, 01 May 2001, Karim A. Nassar wrote: - work on all major platforms, (Does this include Gnome 1.4; anyone using Gnome 1.4 out there ?) - make available all the bugfixes since the last official version (Dia 0.86 has had some serious problems with some menu actions, see: Objects/Align/Equal Distance thread. Are there any *users* out there ?) - contain less bugs/more features than the previous release - silence complaining about bugs, which got fixed month ago - be localized properly ? I think one of the biggest features would be increasing extensibility or providing more and easier to find documentation on extending dia, be it plugins or tools that work along side Dia. I would like to vote for a shapes catalog/plugin scheme. I have tried to insert the Logic shapes into my (.86) install, and it does not work. I have sent this out to the list but recieved no correct answer. It seems that there are others of you that have created shapes and even shape libraries. Maybe a central repository for them? There will definitely be some point where it doesn't make sense to distribute all sheets with Dia, simply because the scroll-down list becomes too long and start-up too slow. We will have to come up with a scheme to handle this. The first step will be to just have a repository with pre-packaged sheets and descriptions as tar-balls. Next would be having Dia be able to connect to there and download them itself. If someone can help me get Logic shapes working, I will help maintain this aspect (provided others agree it is a good idea) :) Let's look at this off-line, and see if we can get it to work. -Lars -- Lars Clausen (http://shasta.cs.uiuc.edu/~lrclause) | Hårdgrim of Numenor I do not agree with a word that you say, but I| Retainer of Sir Kegg will defend to the death your right to say it.| of Westfield --Evelyn Beatrice Hall paraphrasing Voltaire | Chaos Berserker of Khorne
Re: New version soon?
On Tue, 1 May 2001, Dag Wieers wrote: On Tue, 1 May 2001, Karim A Nassar wrote: If someone can help me get Logic shapes working, I will help maintain this aspect (provided others agree it is a good idea) :) If you're willing to send the files, I'll look at it. But tell again what you wanted to do and what fails with your (?) version. I learned a lot by switching between de 0.86 and a CVS-version ;) Well, here we see another reason to release more often: There are many incremental differences between 0.86 and CVS, and we can't easily tell if the Logic shapes depend on some of those. I can't even tell if the version I have installed locally is a pure 0.86 or later. -Lars -- Lars Clausen (http://shasta.cs.uiuc.edu/~lrclause) | Hårdgrim of Numenor I do not agree with a word that you say, but I| Retainer of Sir Kegg will defend to the death your right to say it.| of Westfield --Evelyn Beatrice Hall paraphrasing Voltaire | Chaos Berserker of Khorne
Re: Problem importing custom models...
On Mon, 30 Apr 2001, Ben A. Hetland wrote: Lars Clausen wrote: On Fri, 27 Apr 2001, Karim A. Nassar wrote: Maybe we can link custom-shapes from the FAQ? It's not very big and with this new info could be helpful. No sooner said than done, my lord:) Of course, it'll be a day or two before the FAQ updates on the webpage. The link is in the FAQ, but did you forget to put 'custom-shapes' on your web-site (as linked from a previous message)? (Hmmm... maybe we should just forget that one...) Ah, yes, I forgot that on my website. That was just a temp setup while the main page refreshed. Also the link from Dia's home page (FAQ part) doesn't work properly on my Netscape (under W2k), and I think it's due to the lack of an extension to that file name (NS just wants to save it to a file). IE5.5 handles it, but it seems to have problems with the rest of the web site (extremely slow)... That's because the web-server at lysator is brain-dead and thinks it's an octet-stream rather than text/plain or text/latin-1. Maybe it's an idea to add a '.txt' for those brain-dead environments that think that using a file extension to decide what to do with a file is a good enough solution? At least for the web version... We may have to... -Lars -- Lars Clausen (http://shasta.cs.uiuc.edu/~lrclause) | Hårdgrim of Numenor I do not agree with a word that you say, but I| Retainer of Sir Kegg will defend to the death your right to say it.| of Westfield --Evelyn Beatrice Hall paraphrasing Voltaire | Chaos Berserker of Khorne
Re: New version soon?
On Mon, 30 Apr 2001, Claus Sørensen wrote: What is the roadmap for the next version? Is there a feature freeze? One of the main engines in Open Source/Free Softeware projects is release often so you don't have these huge gaps between releases. Couldn't you release more often so more of us can benefit for the work you have been doing since last release. Then reason we aren't seeing a release is that our maintainer, James Henstridge, seems to have slunk away somewhere. -Lars -- Lars Clausen (http://shasta.cs.uiuc.edu/~lrclause) | Hårdgrim of Numenor I do not agree with a word that you say, but I| Retainer of Sir Kegg will defend to the death your right to say it.| of Westfield --Evelyn Beatrice Hall paraphrasing Voltaire | Chaos Berserker of Khorne
Re: New version soon?
On Mon, 30 Apr 2001, Claus Sørensen wrote: From: Lars Clausen [EMAIL PROTECTED] Then reason we aren't seeing a release is that our maintainer, James Henstridge, seems to have slunk away somewhere. How long have he been away? The last message from him was 4 weeks ago today. Given the amount of activity we're seen recently, one would think he'd be more around. If you are talking about months then it is time for the rest of the developing team to find another maintainer. But first try to reach for him and hear what is going on. I mailed him directly about a week ago (in case his dia-list mail goes somewhere else), but no reply yet. Open Source projects some times lost their maintainers so it is not an unknown situation. The reasons can be many: New job, girl/boyfriend/spouse, lost interest, ill, dead, new religion, travelling around the world, found a more interesting OS-project... Let's hope it some of the more happy reasons but that shouldn't stop the project. Very true. Let's get the standard properties into the rest of the applicable places, and wrap up any other bug we can find. Claus Sorensen K L I D Chairman --Mobile: +45 20 94 62 34 Noddelunden 110 Commercial Linux Email: [EMAIL PROTECTED] DK-2765 Smorum Association of Denmark Web:www.klid.dk Claus, jeg har prøvet at finde frivillige til at oversætte resten af Dia til dansk, specielt de mere specialiserede sheets. Jeg har klaret ca. halvdelen, tror du du kan finde nogen der kender de korrekte fagtermer på dansk? Det er kun ganske lidt arbejde det drejer sig om. -Lars -- Lars Clausen (http://shasta.cs.uiuc.edu/~lrclause) | Hårdgrim of Numenor I do not agree with a word that you say, but I| Retainer of Sir Kegg will defend to the death your right to say it.| of Westfield --Evelyn Beatrice Hall paraphrasing Voltaire | Chaos Berserker of Khorne
Re: Problem importing custom models...
On Fri, 27 Apr 2001, Karim A. Nassar wrote: Maybe we can link custom-shapes from the FAQ? It's not very big and with this new info could be helpful. No sooner said than done, my lord:) Of course, it'll be a day or two before the FAQ updates on the webpage. -Lars -- Lars Clausen (http://shasta.cs.uiuc.edu/~lrclause) | Hårdgrim of Numenor I do not agree with a word that you say, but I| Retainer of Sir Kegg will defend to the death your right to say it.| of Westfield --Evelyn Beatrice Hall paraphrasing Voltaire | Chaos Berserker of Khorne
Re: Problem importing custom models...
On Fri, 27 Apr 2001, Karim A. Nassar wrote: After following the good advice from this thread, I copied the Logic shapes and sheet from the cvs tree into my ~.dia and started dia recieving these messages: *** WARNING **: could not find svg namespace ** WARNING **: could not load shape file /home/kan4/.dia/shapes/Logic/any shape here The location of the shapes/sheet (~/.dia or /usr/share/dia) made no difference -- same error. I looked at the xml for my existing shapes/sheets and the Logic shapes/sheet and could find only these differences: * Existing: uses xpm graphics format, Logic: png I edited the .shape changing png to xpm and converted the images -- no difference Dia understands both. * Some .shapes seems to have different xmlns attributes in the shape tag Examples: from Logic/and.shape: xmlns:svg=http://www.w3.org/2000/svg; from Circuit/ground.shape: xmlns:svg=http://www.w3.org/TR/2000/03/WD-SVG-2303/DTD/svg-2303-stylable.dtd; putting the existing xmlns into the Logic files made no difference. Dia actually ignores this field, it's just there if somebody needs to find the syntax or run it through a checker. It would be easier if you showed us the start of the sheet and shape file. I'm guessing that your naming is wrong. -Lars -- Lars Clausen (http://shasta.cs.uiuc.edu/~lrclause) | Hårdgrim of Numenor I do not agree with a word that you say, but I| Retainer of Sir Kegg will defend to the death your right to say it.| of Westfield --Evelyn Beatrice Hall paraphrasing Voltaire | Chaos Berserker of Khorne
Re: Problem importing custom models...
On Fri, 27 Apr 2001, Karim A. Nassar wrote: After following the good advice from this thread, I copied the Logic shapes and sheet from the cvs tree into my ~.dia and started dia recieving these messages: *** WARNING **: could not find svg namespace ** WARNING **: could not load shape file /home/kan4/.dia/shapes/Logic/any shape here On Fri, 27 Apr 2001, Lars Clausen wrote: It would be easier if you showed us the start of the sheet and shape file. I'm guessing that your naming is wrong. I think it is important to note that the sheets I am trying to use are from the cvs tree. Here are the details: [...] Ah. And you have your shapes in ~/.dia/shapes/Logic and the sheet is ~/.dia/sheets/Logic.sheet? -Lars -- Lars Clausen (http://shasta.cs.uiuc.edu/~lrclause) | Hårdgrim of Numenor I do not agree with a word that you say, but I| Retainer of Sir Kegg will defend to the death your right to say it.| of Westfield --Evelyn Beatrice Hall paraphrasing Voltaire | Chaos Berserker of Khorne
Re: Problem importing custom models...
On Fri, 27 Apr 2001, Cyril ZEKSER wrote: Lars Clausen a écrit : On Fri, 27 Apr 2001, Cyril ZEKSER wrote: I'd like something more clear to me, if it's possible, and apart from using libtool of course ;-) You're right, the custom-shapes doc isn't useful. Let me try to cook up something. I'm anxiously waiting ;-) *tiptaptiptap...* Well, here's the first part: The export shape code now make a reasonable name out of the directory and filename you give it. If you export Test.shape into ~/.dia/shapes/Mine, the name of the shape will be Mine - Test. Docs will follow in the FAQ. -Lars -- Lars Clausen (http://shasta.cs.uiuc.edu/~lrclause) | Hårdgrim of Numenor I do not agree with a word that you say, but I| Retainer of Sir Kegg will defend to the death your right to say it.| of Westfield --Evelyn Beatrice Hall paraphrasing Voltaire | Chaos Berserker of Khorne
Re: Problem importing custom models...
On Fri, 27 Apr 2001, Cyril ZEKSER wrote: Lars Clausen a écrit : On Fri, 27 Apr 2001, Cyril ZEKSER wrote: I'd like something more clear to me, if it's possible, and apart from using libtool of course ;-) You're right, the custom-shapes doc isn't useful. Let me try to cook up something. I'm anxiously waiting ;-) *tiptaptiptap...* Updated FAQ now in dia-web cvs, also (until the regular FAQ page updates) available at URL:http://shasta.cs.uiuc.edu/~lrclause/faq.html. Hopefully that answers your questions (note that this is for the very newest CVS version, where the shape name was correctly generated). Comments most welcome. -Lars -- Lars Clausen (http://shasta.cs.uiuc.edu/~lrclause) | Hårdgrim of Numenor I do not agree with a word that you say, but I| Retainer of Sir Kegg will defend to the death your right to say it.| of Westfield --Evelyn Beatrice Hall paraphrasing Voltaire | Chaos Berserker of Khorne
Re: ./autogen.sh problems
On Thu, 26 Apr 2001, Karim A. Nassar wrote: I am getting funny errors when I try to configure dia for compilation. I grabbed the cvs today, ran ./autogen and got the following: I am going to run ./configure with no arguments - if you wish to pass any to it, please specify them on the ./autogen.sh command line. Running gettextize... Ignore non-fatal messages. sed: can't read aclocal.m4: No such file or directory You should update your own `aclocal.m4' by adding the necessary macro packages gettext.m4, lcmessage.m4 and progtest.m4 from the directory `/usr/share/aclocal' I have no dir /usr/shar/aclocal and have no idea what this is refering to. I assume that this is what causes the rest of the errors (the balnace of this message). Any suggestions? Do you have all the development packages installed? Look at the Dia homepage for the required packages. Notice that it's not enough to have the library, you need the development stuff as well (-dev packages nder Debian, I think -devel under RedHat). -Lars -- Lars Clausen (http://shasta.cs.uiuc.edu/~lrclause) | Hårdgrim of Numenor I do not agree with a word that you say, but I| Retainer of Sir Kegg will defend to the death your right to say it.| of Westfield --Evelyn Beatrice Hall paraphrasing Voltaire | Chaos Berserker of Khorne
Re: EML plug-in for Dia (fwd)
On Thu, 19 Apr 2001, Alexander Larsson wrote: Can someone take a look at this and commit it to cvs. I'm very short on time currently. -- Forwarded message -- Date: Wed, 18 Apr 2001 23:45:11 +0600 From: Vladimir Sekissov [EMAIL PROTECTED] To: [EMAIL PROTECTED] Subject: EML plug-in for Dia Dear, Alexander. I posted this message to James Henstridge three weeks ago but haven't any response yet. I wrote Event Modeling Language plug-in for Dia. Would you get it from http://disney.surnet.ru/dia and put into CVS if code is correct or/and send me comments and suggestions. Thank you, it looks good. It's in CVS now. I have a bunch of suggestions, but no showstoppers: For the Process object: Make it so that at least some of the text can be written directly into the diagram, rather than having to open the properties dialog each time. I'm not sure which would be more appropriate, as I don't know the EML. Could the first two tabs be combined into one, so there isn't so much dead space? Possibly with a labelled box around them. The Interfaces area doesn't really need the 'Move Up' and 'Move Down' buttons, as they do the same as when you click on the (scrollable) list. My first thought was that they rearranged the interfaces. The interface name entry box should understand Enter. For the three other objects: Use the standard properties code rather than putting GTK code into the objects. It's easier and more future-proof. For the Instantiation object: Make the segment with the arrows on it long enough to have the polyline go past the arrows. For everything: It would be nice if it understood at least the color selection. Many people have asked for UML with color, and it does color now, though not via the standard color selection:( None of these are showstoppers, just ideas for improvement. Thank you for the contribution! -Lars -- Lars Clausen (http://shasta.cs.uiuc.edu/~lrclause) | Hrdgrim of Numenor "I do not agree with a word that you say, but I| Retainer of Sir Kegg will defend to the death your right to say it."| of Westfield --Evelyn Beatrice Hall paraphrasing Voltaire | Chaos Berserker of Khorne
Re: pasted image upside down when printed
On Thu, 19 Apr 2001, Victor Schan wrote: Hi! I imported a gif picture and when printed out it is upside down. The rest of the diagramm is ok, only the pictures I pasted in. Is this a bug or have I done anything wrong? Which version of Dia are you using? An early version of the image handling did this, but I thought we had it out now. -Lars -- Lars Clausen (http://shasta.cs.uiuc.edu/~lrclause) | Hrdgrim of Numenor "I do not agree with a word that you say, but I| Retainer of Sir Kegg will defend to the death your right to say it."| of Westfield --Evelyn Beatrice Hall paraphrasing Voltaire | Chaos Berserker of Khorne
Re: Question about dia gnome
On Mon, 16 Apr 2001, adam morley wrote: I was wondering, GNOME doesn't seem to have any diagramming software, and dia seems to use a lot of gnome interfaces (glib, gtk, imlib, etc.) are there any plans to integrate dia with gnome? We want to keep Dia from depending on Gnome, but still use as much of Gnome as we can. Dia is listed in the Gnome software list, under productivity (this obviously doesn't apply to the developers:). If it doesn't come with the rest of Gnome when installed, that would be the fault of the makers of the distribution you use. I know it comes with Redhat. -Lars -- Lars Clausen (http://shasta.cs.uiuc.edu/~lrclause) | Hrdgrim of Numenor "I do not agree with a word that you say, but I| Retainer of Sir Kegg will defend to the death your right to say it."| of Westfield --Evelyn Beatrice Hall paraphrasing Voltaire | Chaos Berserker of Khorne
Re: Target Gtk+2.0 (was Re: plug-ins/renderer.inc)
On Sun, 15 Apr 2001, Cyrille Chepelov wrote: Le jeu, avr 12, 2001, 09:10:31 +0200, Hans Breuer a crit: The win32 port is almost up-to-date (my last commit is about a week ago, next one probably this weekend). I suggested a Dia branch "Target Gtk+2.0" in a mail to the list at Fri, 23 Feb 2001 16:06:34 +0100 but there was no consensus about the right time to it in the thread. I really don't see the point in targetting Gtk 2 until gtk2 is in (sort of) "release candidate" stage ; until major distributors ship libgtk2, I fear it's quite a waste of effort to work on two trees. Of course, if gtk2 is really that close already, then it makes sense to start the fork ASAP (but we should probably concentrate on a really clean 0.87, and a fully debugged .88 just afterwards, say, no more than two months after 0.87. I really think we don't have enough releases, and that we keep answering to FAQs in a "sorry, it's fixed in CVS for months but we haven't released anything" style.) Announcing a .88 just after .87 (while keeping the lid relatively closed on new features) would be an excellent way to test the new code (since .86) against real bugs, and would give translators a really nice time frame. I agree on this; aiming for a fast semi-stable release would give us time to then do the more serious improvements (better fonts, standard properties, etc) and work (slowly) towards a 1.0 release. I prefer to have a quality last version out while while we do major reconstructions. Maybe we should call it 0.90 just to mark its emphasis on stability? -Lars -- Lars Clausen (http://shasta.cs.uiuc.edu/~lrclause) | Hrdgrim of Numenor "I do not agree with a word that you say, but I| Retainer of Sir Kegg will defend to the death your right to say it."| of Westfield --Evelyn Beatrice Hall paraphrasing Voltaire | Chaos Berserker of Khorne
Re: Target Gtk+2.0 (was Re: plug-ins/renderer.inc)
On Sun, 15 Apr 2001, Cyrille Chepelov wrote: Le dim, avr 15, 2001, 10:16:30 -0500, Lars Clausen a crit: I agree on this; aiming for a fast semi-stable release would give us time to then do the more serious improvements (better fonts, standard properties, etc) and work (slowly) towards a 1.0 release. I prefer to have a quality last version out while while we do major reconstructions. Maybe we should call it 0.90 just to mark its emphasis on stability? For the version beyond 0.87, probably. I second this. By the way, has anyone got news of James recently ? Nope, nary a word. Until we do, we should just compile a list of what we want from 0.87. My main points: Translate all simple properties dialogs into standard properties. Use auto-generated defaults dialogs where possible. Correct bounding boxes for all elements. If somebody can figure out how to do the postscript: Real font support. -Lars -- Lars Clausen (http://shasta.cs.uiuc.edu/~lrclause) | Hrdgrim of Numenor "I do not agree with a word that you say, but I| Retainer of Sir Kegg will defend to the death your right to say it."| of Westfield --Evelyn Beatrice Hall paraphrasing Voltaire | Chaos Berserker of Khorne
Re: FS patch
On Fri, 13 Apr 2001, David C. Thompson wrote: I know that it may be too late for dia 0.87, but I have a patch for the FS objects. It also touches app/disp_callbacks.c in one small spot to allow Object menus to have submenus. It should not affect any other objects. The patch is about 24k uncompressed, so I've put it on the web rather than attach it. It's at http://madlab.me.utexas.edu/~dcthomp/dload/dia-FS-010413-1.patch.gz The Changelog/CVS entry would be - * app/disp_callbacks.c: - Allow submenus in DiaMenus. - Don't require DiaMenu.title to be non-null (so that submenus aren't required to have titles) * objects/FS/function.c: - object menu now has submenus, an almost complete list of standard-basis verbs, and a start on nouns. - box padding, dash length, and other visual attributes are scaled by the font size... eventually font height will be a function property. - draw functions with smaller font to make initial view hold a reasonable number of functions. - This is the second time I've submitted it. If there's something wrong with it, please let me know so I can fix it. Seems to work. That sure is a lot of words in those menus. Now I don't know the FS system, but a submenu link Verb/Channel/Guide/Guide/Guide seems funny to me. Is this a standardized setup? Also, it would be nice if the two Fn entries were togglemenus. It will be in CVS in the morning. Just so that we have somebody do it, I will make sure to apply any patch sent to the list, unless somebody else objects to it. I'll give one day for objections, and send an acknowledgement when it's in CVS. -Lars -- Lars Clausen (http://shasta.cs.uiuc.edu/~lrclause) | Hrdgrim of Numenor "I do not agree with a word that you say, but I| Retainer of Sir Kegg will defend to the death your right to say it."| of Westfield --Evelyn Beatrice Hall paraphrasing Voltaire | Chaos Berserker of Khorne
Re: dia png and HTML Maps (fwd)
On Thu, 12 Apr 2001, Alexander Larsson wrote: Forwarding to list. / Alex -- Forwarded message -- Date: Tue, 10 Apr 2001 21:26:35 +0200 (MET DST) From: poete stephane [EMAIL PROTECTED] To: [EMAIL PROTECTED] Subject: dia png and HTML Maps Using dia, I usually export dia figures in png, to insert the png figures in html documents. I need, for some of these diagrams, to build HTML maps, so I tried to understand how the native XML dia is converted to png. I did not open any dia source code, because I wanted to write some python to build map for the dia objects I'm interested in. If there is a way to understand the png export without reading dia C code, please let me know. That's a tricky question. Rendering is done at a fairly low level. The PNG renderer only knows about boxes, lines, text strings and such, it never knows that, say, a box, a line and a text makes up a class. The objects don't know what they're being rendered as, they just draw themselves as lines, boxes etc. However, it should be possible to figure out the extents of the PNG image from the XML file. Unzip a .dia file and have a look. -Lars -- Lars Clausen (http://shasta.cs.uiuc.edu/~lrclause) | Hrdgrim of Numenor "I do not agree with a word that you say, but I| Retainer of Sir Kegg will defend to the death your right to say it."| of Westfield --Evelyn Beatrice Hall paraphrasing Voltaire | Chaos Berserker of Khorne
Re: bug?
On Thu, 12 Apr 2001, Alexander Larsson wrote: On 5 Apr 2001, Lars Clausen wrote: So it fails in image.c, line 901, which is g_free(diafile_dir); diafile_dir is created by get_directory(). Now, Alex, would you explain why get_directory doesn't use the standard g_dirname to get the directory part of a filename? :) I claim temporary insanity. Accepted. You will only get 5 years in the Gulag, then:) -Lars -- Lars Clausen (http://shasta.cs.uiuc.edu/~lrclause) | Hrdgrim of Numenor "I do not agree with a word that you say, but I| Retainer of Sir Kegg will defend to the death your right to say it."| of Westfield --Evelyn Beatrice Hall paraphrasing Voltaire | Chaos Berserker of Khorne
Re: need some help =)
On Thu, 12 Apr 2001, B. C. J. O. wrote: I apologise for my first post being a plea for help, but I'm having a problem /w Dia that I can't seem to solve.. it's in the realm of those things that seem to point at brane-damage in the operator, but I've fought with it for a couple of days now and as I feel the mouse related RSI starting to kick in, I thought I'd look elsewhere for aid. I have been building a sizable diagram describing the navigation path for a web site, and there are a number of basic functions that seem very well hidden. I can't seem to set the default font for drawings. Everything defaults to courier. When I select a group of objects, the properties dialog doesn't set the properties for that group -- only for the first of the objects selected in the tree. =/ Is there another mechanism to do this? I have potentially multiple hundreds of objects in this drawing, and if I have to mouse around to open the preferences dialog for every one, I'm going to have to amputate my hand. Any help greatfully appreciated. =) Fortunately, help is on the way! To set the default settings for new objects, double-click on the object icon, in this case either the text icon or the icon of whichever object you're using. Some objects don't allow setting the font, though, and some don't allow the defaults to be set. As for setting properties for a group of objects, you just need to Group Objects. Then you can set any properties (provided the objects use the standard property system). -Lars -- Lars Clausen (http://shasta.cs.uiuc.edu/~lrclause) | Hrdgrim of Numenor "I do not agree with a word that you say, but I| Retainer of Sir Kegg will defend to the death your right to say it."| of Westfield --Evelyn Beatrice Hall paraphrasing Voltaire | Chaos Berserker of Khorne
Re: EPS export and handles on BezierLine
On Tue, 10 Apr 2001, Richard Sharpe wrote: Hi, It seems that the handles on a bezier line define the actual bounding box produced when you export a DIA diagram as an EPS ... This can cause ugly whitespace around the diagram. Can this be easily fixed? Not really easily -- we'd need to have a separate bounding box calculation for displays that shouldn't display the control handles, and getting the BB right for a bezier line is not quite trivial. We do have some code we're trying out now. -Lars -- Lars Clausen (http://shasta.cs.uiuc.edu/~lrclause) | Hrdgrim of Numenor "I do not agree with a word that you say, but I| Retainer of Sir Kegg will defend to the death your right to say it."| of Westfield --Evelyn Beatrice Hall paraphrasing Voltaire | Chaos Berserker of Khorne
Re: bug?
On Thu, 05 Apr 2001, Ben A. Hetland wrote: Sebastian Fischmeister wrote: Anyway, Sebastian, try this patch: It works; Lars you rock! Yes, but assuming the directory separator is '/'... which of course any good OS accepts, and even some of the poor ones accept it internally ;-) If Glib had a function to tell me the seperator, I'd use that. (Ah yes... and assuming it _has_ directories, of course!) Well, we're assuming *that* all over the install code. -Lars -- Lars Clausen (http://shasta.cs.uiuc.edu/~lrclause) | Hrdgrim of Numenor "I do not agree with a word that you say, but I| Retainer of Sir Kegg will defend to the death your right to say it."| of Westfield --Evelyn Beatrice Hall paraphrasing Voltaire | Chaos Berserker of Khorne
Re: bug?
On Thu, 05 Apr 2001, David C. Thompson wrote: Yes, but assuming the directory separator is '/'... which of course any good OS accepts, and even some of the poor ones accept it internally ;-) If Glib had a function to tell me the seperator, I'd use that. It has a macro... is that good enough? http://developer.gnome.org/doc/API/glib/glib-standard-macros.html#G-DIR-SEPARATOR-CAPS Thank you! Didn't expect it there. Fix is in CVS. But then again, looking at lib/intl.c and intl/*.c, there's quite a few hardcoded filenames that use '/'. -Lars -- Lars Clausen (http://shasta.cs.uiuc.edu/~lrclause) | Hrdgrim of Numenor "I do not agree with a word that you say, but I| Retainer of Sir Kegg will defend to the death your right to say it."| of Westfield --Evelyn Beatrice Hall paraphrasing Voltaire | Chaos Berserker of Khorne
Re: Creating images
On Wed, 4 Apr 2001, Justin Noack wrote: Hello, I made a very nice table lay out with Dia and it is very nice. But I need to be able to use what I made on the net like a gif or jpg. How can I convert the dia or export the file to where I can use it on the webpage? Use the File-Export menu item from the right button menu, and select PNG. If you're worried about browsers not supporting PNG, a program like convert (from the ImageMagick package) can do conversions to GIF (Dia can't because of patent issues). Definitely don't use JPEG, as you'll get a lot of artifacts around the white/black borders. -Lars -- Lars Clausen (http://shasta.cs.uiuc.edu/~lrclause) | Hrdgrim of Numenor "I do not agree with a word that you say, but I| Retainer of Sir Kegg will defend to the death your right to say it."| of Westfield --Evelyn Beatrice Hall paraphrasing Voltaire | Chaos Berserker of Khorne
Re: Logic Gates
On Tue, 03 Apr 2001, Michael Luchtan wrote: Hello all I'm currently looking for a program where I can lay out some simple boolean circuits for discussion purposes, and it seems like Dia would work good. I can't find any logic gates (and, or, not) any where, and would be willing to make some if none are already around, and if I could get some help getting started. Is it as simple as building the shape that you want in Dia and saving it in an appropriate way? Any suggestions on where to start would be appreciated. Thanks! I made a set (as a first try at making shapes) which is now included in the CVS version. They can probably be improved upon, so feel free to mess with them. Making new shapes is almost just building the shape in Dia and then saving it as .shape. You'll have to set up a sheet for it, and it probably needs tweaking, but not much. -Lars -- Lars Clausen (http://shasta.cs.uiuc.edu/~lrclause) | Hrdgrim of Numenor "I do not agree with a word that you say, but I| Retainer of Sir Kegg will defend to the death your right to say it."| of Westfield --Evelyn Beatrice Hall paraphrasing Voltaire | Chaos Berserker of Khorne
GRAFCET steps
I was about to make the GRAFCET objects use standard properties, when I noticed very strange behaviour for several objects. The 'Step' object doesn't understand moving the handles (only the up-left part seems to work) and doesn't resize or even do BB to account for large texts, the 'Condition' object is empty except for a text, and the 'Transition' object behaves similarly to the 'Step' object. Cyrille, I see your name in the files. Do you have anything to say in your defense? :) -Lars -- Lars Clausen (http://shasta.cs.uiuc.edu/~lrclause) | Hrdgrim of Numenor "I do not agree with a word that you say, but I| Retainer of Sir Kegg will defend to the death your right to say it."| of Westfield --Evelyn Beatrice Hall paraphrasing Voltaire | Chaos Berserker of Khorne
Re: DocBook suggestions and new spec-file
On Mon, 2 Apr 2001, Cyrille Chepelov wrote: Le lun, avr 02, 2001, 08:45:05 +0200, Dag Wieers a crit: Why not answer this myself. heh -- I should read the whole ML before answering... Unfortunately, these packages are called PyXML and that's what's causing all the confusion. Maybe the error-message about python and python-xml should clarify this by saying python-xml or PyXML ? Maybe even a url ? Yes, definitely. I'm not at home, and the bandwith sucks here, so feel free to commit this before me. Ah, Cyrille, he can't. He doesn't have CVS write access. -Lars -- Lars Clausen (http://shasta.cs.uiuc.edu/~lrclause) | Hrdgrim of Numenor "I do not agree with a word that you say, but I| Retainer of Sir Kegg will defend to the death your right to say it."| of Westfield --Evelyn Beatrice Hall paraphrasing Voltaire | Chaos Berserker of Khorne
Re: Network shapes and sample network diagram
On Tue, 3 Apr 2001, Dag Wieers wrote: On Mon, 2 Apr 2001, Cyrille Chepelov wrote: Indeed, we have to begin thinking seriously about some kind of compat layer, to put old objects to rest. I've renamed "GRAFCET - Vector" into "GRAFCET - Arc" (for the moment, the code is the same), I'd really like to kill RenderObjects and make them all shapes (less code more good), and we have those occasional XML namespace problems. This fits nicely in Lars suggestion to have some sort of Sheet browser/editor. If a diagram file is loaded with shapes that does not exist, one might want to browse the Sheet browser to pick a (matching) shape. (And by showing the old name, one can easily match it with a new one). This way we have more freedom to change the namespaces (of SDL and such) without having to worry about backward-compatability. Another nice addition to this is to have a preview-shot of the shape that is selected (while browsing). I think the idea has merit, but I'll focus on the .087 release for now. Coming up with an interface for a browser would be a useful pastime, though. And what about (on-the-fly) creation of small icons of shapes (instead of the xpm-files) now we're at it. :) Well, saving as shape file gives an icon as well, but the quality is not very high. It should only be a last resort. -Lars -- Lars Clausen (http://shasta.cs.uiuc.edu/~lrclause) | Hrdgrim of Numenor "I do not agree with a word that you say, but I| Retainer of Sir Kegg will defend to the death your right to say it."| of Westfield --Evelyn Beatrice Hall paraphrasing Voltaire | Chaos Berserker of Khorne
Re: Core dump on dia file load (fwd)
On Sun, 1 Apr 2001, Cyrille Chepelov wrote: It doesn't core dump on me (CVS head); however, dia behaves strangely when I load it: no way to use the horizontal scrollbars, the scroll tool doesn't work either. After a few manipulations, the canvas becomes black. dia survives a few open/close/open/close cycles (precisely, I haven't been able to crash it). Looking at the file, it looks like there are a few incorrect values in the beziergons: point val="6.79704e-313,8.34416e-308"/ and later: point val="6.79704e-313,8.34416e-308"/ and again later: point val="6.79704e-313,8.34416e-308"/ [...] I believe I put a patch in there at one time that upped the version number on the beziergon saves and used that to make sure that old diagrams were loaded correctly as well as new. Did that get lost in the underflow? -Lars -- Lars Clausen (http://shasta.cs.uiuc.edu/~lrclause) | Hrdgrim of Numenor "I do not agree with a word that you say, but I| Retainer of Sir Kegg will defend to the death your right to say it."| of Westfield --Evelyn Beatrice Hall paraphrasing Voltaire | Chaos Berserker of Khorne
How does flowchart/box.c update when text is inserted
I'm looking at bug #51222 (connected lines don't move when text is added). The flowchart/box object nicely shows this behaviour, but I can't figure out how the box outline itself is updated when text is typed. Where is the connection back? Whereever it is, it should call box_update_data. -Lars -- Lars Clausen (http://shasta.cs.uiuc.edu/~lrclause) | Hrdgrim of Numenor "I do not agree with a word that you say, but I| Retainer of Sir Kegg will defend to the death your right to say it."| of Westfield --Evelyn Beatrice Hall paraphrasing Voltaire | Chaos Berserker of Khorne
Re: More fixing (#50903 and #51479)
On Sun, 1 Apr 2001, Cyrille Chepelov wrote: Le dim, avr 01, 2001, 02:45:57 -0500, Lars Clausen a crit: I've committed fixes for bugs #50903 and 51479. For bug #51233, I need advice from someone with more knowledge about beziers: Is there an easy way to find the true bounding box of the bezier? For page fitting purposes, it seems we use some polyline approzimation, but isn't there a way to find the maximum extents of the bezier? it's defined to be bounded by its controling polygon. If that's what it uses now, that does not work properly. Try making a bezierline, make the page fit to 1x1, and drag the control points outwards (to make a C-like shape). The bezierline goes outside the page, though the page is resized to large than the control points. -Lars -- Lars Clausen (http://shasta.cs.uiuc.edu/~lrclause) | Hrdgrim of Numenor "I do not agree with a word that you say, but I| Retainer of Sir Kegg will defend to the death your right to say it."| of Westfield --Evelyn Beatrice Hall paraphrasing Voltaire | Chaos Berserker of Khorne
Re: More fixing (#50903 and #51479)
On Mon, 2 Apr 2001, James Henstridge wrote: On 1 Apr 2001, Lars Clausen wrote: On Sun, 1 Apr 2001, Cyrille Chepelov wrote: Le dim, avr 01, 2001, 02:45:57 -0500, Lars Clausen a crit: I've committed fixes for bugs #50903 and 51479. For bug #51233, I need advice from someone with more knowledge about beziers: Is there an easy way to find the true bounding box of the bezier? For page fitting purposes, it seems we use some polyline approzimation, but isn't there a way to find the maximum extents of the bezier? it's defined to be bounded by its controling polygon. If that's what it uses now, that does not work properly. Try making a bezierline, make the page fit to 1x1, and drag the control points outwards (to make a C-like shape). The bezierline goes outside the page, though the page is resized to large than the control points. Inside lib/geometry.c, there is some code for approximating beziers. It may not be adequate for bounding box calculations, as it would probably end up missing some of the curve (and I am not sure how much correction would be appropriate). Good morning James:) I would hope it could do enough to keep the page to fit. In other news, could you set up us the bomb^W^W^Wa plan for 0.87? A fast mostly-bugfix release would be good to include with the new Gnome stuff. Possibly the new doc stuff shouldn't be included right now (or be in a seperate CVS project?), Dag Wieers reports trouble compiling it, and right before a release is not quite the time to insert a new big part. I think a dia-doc project would be good. -Lars -- Lars Clausen (http://shasta.cs.uiuc.edu/~lrclause) | Hrdgrim of Numenor "I do not agree with a word that you say, but I| Retainer of Sir Kegg will defend to the death your right to say it."| of Westfield --Evelyn Beatrice Hall paraphrasing Voltaire | Chaos Berserker of Khorne
Shapes don't fall on grid boundaries
For someone more accustomed with the custom shapes code, there is something that needs to be fixed wrt to the coord offset. Try using the Network/Bigtower PC with Snap to Grid on. It doesn't fall on the grid at all until you explicitly move it, a nasty surprise for those who want their things lined up quickly. Maybe the offset need to just be subtracted at shape load time and never touched again? Maybe even the same with the scaling? -Lars -- Lars Clausen (http://shasta.cs.uiuc.edu/~lrclause) | Hrdgrim of Numenor "I do not agree with a word that you say, but I| Retainer of Sir Kegg will defend to the death your right to say it."| of Westfield --Evelyn Beatrice Hall paraphrasing Voltaire | Chaos Berserker of Khorne
New name for 'Reset tools after create'
Dag Wieer [EMAIL PROTECTED] managed to miss the 'Reset tools after create' for a very long time. We discussed the naming, and I came up with 'Reset to pointer after create'. Do you all think this is a good enough change to be worth messing with peoples minds and translations? Ok, I'm done spamming you for the night. I hope we can get to 0.87 quickly. I almost made an Aprils Fools joke with you, but didn't have the time: Claiming to have fixed the font support, and including a reasonably-looking patch that would change the font list to just say 'Aprils fool'. -Lars -- Lars Clausen (http://shasta.cs.uiuc.edu/~lrclause) | Hrdgrim of Numenor "I do not agree with a word that you say, but I| Retainer of Sir Kegg will defend to the death your right to say it."| of Westfield --Evelyn Beatrice Hall paraphrasing Voltaire | Chaos Berserker of Khorne
Re: Compiling from CVS source
On Sat, 31 Mar 2001, James K. Lowden wrote: Lars, Mark, Ben: Thanks! I moved/linked /usr/local to /usr, symlinked gdk-pixbuf.m4, updated my $PATH, and created LD_LIBRARY_PATH. Autogen then complained about libxml-1.7 something being too old, so I installed libxml-1.8.11-1. Then it worked. Not without warnings, but OK. I got far enough along in the make to exhaust available disk space, though. (I guess that 30GB that's in the mail is going to come in handy.) I just bought Havoc Pennington's "GTK+/Gnome Application Development" http://www.newriders.com/books/title.cfm?isbn=0735700788 (full source also online!) I guess I have my reading to do while UPS is en route. One more question in passing. I'm intending to use Dia in Gnome. The X client (dia) is linux; the X server is xfree86 on NetBSD. Autogen ended with: Gnome support: no Build bonobo component (not supported): no Gnome-print support (not recommended): no Should those all be turned off in my case? You can use it in Gnome with them turned off, but you might want to do --enable-gnome to get some extra Gnome functionality. Don't use Gnome print yet. -Lars -- Lars Clausen (http://shasta.cs.uiuc.edu/~lrclause) | Hrdgrim of Numenor "I do not agree with a word that you say, but I| Retainer of Sir Kegg will defend to the death your right to say it."| of Westfield --Evelyn Beatrice Hall paraphrasing Voltaire | Chaos Berserker of Khorne
Re: [RFC] changing a little the arrow interface
On Sat, 31 Mar 2001, Ben A. Hetland wrote: [...] I have also observed what I think is a different problem, but still closely related to the arrow heads. When an arrow is pointing into another shape -- say connected to the left side of a rectangle, for instance -- then the tip of the arrow "penetrates" the other object's line (part of the tip is inside the rectangle in this case). This is especially noticeable with larger line widths, and I think it looks very ugly that way... Which is exactly why this isn't a 0.87 problem. We need to redesign how arrows are drawn completely. I was thinking of having the arrow be more of a part of the line, rather than something the object slaps on. Then the line-drawing code can do the arrowhead in the *right* way, and account for that in the line. But right now we need to look at what needs to be done for 0.87. I'm afraid that better font support would be as much work as better arrowheads, but I would really like the RLE stuff to get working. Do we want as many objects as possible converted to standard properties? For the FS objects, is there any reason to have a big property entry for the text field, or can I just kill that? -Lars -- Lars Clausen (http://shasta.cs.uiuc.edu/~lrclause) | Hrdgrim of Numenor "I do not agree with a word that you say, but I| Retainer of Sir Kegg will defend to the death your right to say it."| of Westfield --Evelyn Beatrice Hall paraphrasing Voltaire | Chaos Berserker of Khorne
Re: Compiling from CVS source
On Fri, 30 Mar 2001, Ben A. Hetland wrote: "James K. Lowden" wrote: I'm new to the Gnu development environment, and I'm hoping someone can tell me what obvious thing I need to do to build dia. [...] Turbolinux came with all the developer's tools (except gdk-pixbuf, which I think I built and installed correctly to /usr/local/lib from ftp.rpmfind.net/linux/gnome.org/unstable/sources/gdk-pixbuf). I didn't get the autogen-stuff to work with the libraries/headers stored under /usr/local/... Got about the same errors as you get. The "quick-and-dirty" solution for me was to install these (self-built) libraries under /usr/include and /usr/lib directly. I didn't investigate further, but there seems to be some issues here that the concept of /usr/local is not quite understood by the "automatic" tools (maybe it's as simple as an env. variable or rc-file, though...). It should be that simple. Make sure /usr/local/bin is first in your PATH, and that /usr/local/lib is first in your LD_LIBRARY_PATH. That's all it should need. -Lars -- Lars Clausen (http://shasta.cs.uiuc.edu/~lrclause) | Hrdgrim of Numenor "I do not agree with a word that you say, but I| Retainer of Sir Kegg will defend to the death your right to say it."| of Westfield --Evelyn Beatrice Hall paraphrasing Voltaire | Chaos Berserker of Khorne
What to do before next release?
I think I heard some ruminations about an impending release 0.87. I think it'd be good to have a new release soon, to go into distros together with Gnome 1.4. So what do we want to fix before then? We should not add any new features right now, the question is how much do we want to fix? Things like RLE encoding of images and font handling springs to mind as things that really need to be fixed but would take some work. James, do you have a roadplan? -Lars -- Lars Clausen (http://shasta.cs.uiuc.edu/~lrclause) | Hrdgrim of Numenor "I do not agree with a word that you say, but I| Retainer of Sir Kegg will defend to the death your right to say it."| of Westfield --Evelyn Beatrice Hall paraphrasing Voltaire | Chaos Berserker of Khorne
Re: [RFC] changing a little the arrow interface
On Sat, 31 Mar 2001, Cyrille Chepelov wrote: Hi again, I'd like to make a "little" change to the arrow interface, most probably after 0.87: void arrow_draw(Renderer *renderer, ArrowType type, -Point *to, Point *from, +Point *to, const Point *from, real length, real width, real linewidth, -Color *fg_color, Color *bg_color); +const Color *fg_color, const Color *bg_color); Specifically, I'd like to make "to" mutable *and* to have all individual drawing code modify it, so that the caller knows where to end its [poly]lines. This should rid us of those disgracious artefacts we can see at the ends of filled triangle arrows (and other types too, probably). This change will require sweeping through all objects, so I won't do this if .87 is a couple weeks away from us ; but as soon as it's released, it gets high on my TODO list. Comments ? I've been thinking about the whole arrow issue, and I didn't come up with anything that easy. If it can really solve all the problems with curves and arrows, I'd be very happy. -Lars -- Lars Clausen (http://shasta.cs.uiuc.edu/~lrclause) | Hrdgrim of Numenor "I do not agree with a word that you say, but I| Retainer of Sir Kegg will defend to the death your right to say it."| of Westfield --Evelyn Beatrice Hall paraphrasing Voltaire | Chaos Berserker of Khorne
Multiple text fields in a shape?
My officemate suggested allowing multiple object fields in a shape, using TAB to cycle between them. Some objects with multiple pieces of text could also use that. I think it sounds like a good idea, but it'll need some infrastructure (certainly not for the upcoming release). Any problems with this? -Lars -- Lars Clausen (http://shasta.cs.uiuc.edu/~lrclause) | Hrdgrim of Numenor "I do not agree with a word that you say, but I| Retainer of Sir Kegg will defend to the death your right to say it."| of Westfield --Evelyn Beatrice Hall paraphrasing Voltaire | Chaos Berserker of Khorne
Re: Multiple text fields in a shape?
On Thu, 22 Mar 2001, Hubert Figuiere wrote: On jeudi, mars 22, 2001, at 08:25 , Lars Clausen wrote: My officemate suggested allowing multiple object fields in a shape, using TAB to cycle between them. Some objects with multiple pieces of text could also use that. I think it sounds like a good idea, but it'll need some infrastructure (certainly not for the upcoming release). Any problems with this? The same way, I have thought of multiple shapes for an object. For example you could have a "computer" object and a choice of different shapes for different models. Since they are supposed to have the same diagram properties it seems to be logical. Well, these fields (text fields was what I meant) would all be visible at the same time. He's doing lambda diagrams, where he'd want to do something like _ / \ B-A \ \ / \ / \ / \ A-B Both of these text fields should be a normal text object (the Lambda should be a read-only text object, but we can't get the Symbol font to print correctly) that can be edited seperately. Since we're not using the TAB key for anything, I think this would be useful. Also useful for things like UML Small Package or Component that contains several text fields already, but where you have to open Properties to edit them. Problem is that the object would require to get the keypresses, or somehow tell the system about the multiple text fields. -Lars -- Lars Clausen (http://shasta.cs.uiuc.edu/~lrclause) | Hrdgrim of Numenor "I do not agree with a word that you say, but I| Retainer of Sir Kegg will defend to the death your right to say it."| of Westfield --Evelyn Beatrice Hall paraphrasing Voltaire | Chaos Berserker of Khorne
Re: config.h again
On Wed, 21 Mar 2001, Ben A. Hetland wrote: "Ben A. Hetland" wrote: Lars Clausen wrote: Despite Hans' Changelog entry, there are still many files without config.h included: [...snip...] [ben@chpc204 dia]$ find . -name \*.c | xargs grep '#include *["]config.h["]' | wc -l 142 Ooops, even better: [ben@chpc204 dia]$ find . -name \*.c | xargs grep 'include *["]config.h["]' | wc -l 158 bash-2.03$ find . -name \*.c | xargs grep \ '# *include *["]config\.h["]' | wc -l 161 It's a bloody mess. Is it supposed to be config.h or "config.h"? And we're still missing 67 of them. -Lars -- Lars Clausen (http://shasta.cs.uiuc.edu/~lrclause) | Hrdgrim of Numenor "I do not agree with a word that you say, but I| Retainer of Sir Kegg will defend to the death your right to say it."| of Westfield --Evelyn Beatrice Hall paraphrasing Voltaire | Chaos Berserker of Khorne
Re: config.h again
On Wed, 21 Mar 2001, Ben A. Hetland wrote: Lars Clausen wrote: bash-2.03$ find . -name \*.c | xargs grep \ '# *include *["]config\.h["]' | wc -l 161 We could of course also include the TAB in case that's used in some of the files, but this still doesn't check that the inclusion is done _first_, and besides... I checked for the tabs (or rather, I checked that I get the same without including the #). I'll have to think a bit about the firstness... there, only 8 files have that: ./app/render_gnomeprint.c ./app/render_dps_helper.c ./lib/font.c ./lib/charconv.c ./lib/ps ./objects/ER/entity.c ./objects/UML/class_dialog.c ./objects/standard/polycurve.c I can fix those by hand. It's a bloody mess. Is it supposed to be config.h or "config.h"? And we're still missing 67 of them. ...this still doesn't make much difference to your main point here, anyway. And we're down to only 67 now, which should be easy to locate... In fact, I made a three-line shell script that fixed them all. According to my ancient "ABC" on C that happened to be on my bookshelf nearby right now (I program mostly in C++ now, so C doc is sparse...), I "refreshed" my knowledge a little on this subject. It says on the subject vs "": "" A search for the file is made first in the current directory and then in the standard places. the preprocessor looks for the file only in the standard places and not in the current directory. ...but the standard might have updated itself many times after that ;-) Interesting thing is that automake'd makefiles have -I., so they're actually the same. It's more a question of style, but However, this suggests that "" is the most inclusive form (and allows for even local versions in the sub-dirs, which might not be a good idea), so maybe this should be chosen? ...it might be good to show that it's not a per-directory include. On the other hand, I usually see as meaning system includes instead of includes that come with the program. James? -Lars -- Lars Clausen (http://shasta.cs.uiuc.edu/~lrclause) | Hrdgrim of Numenor "I do not agree with a word that you say, but I| Retainer of Sir Kegg will defend to the death your right to say it."| of Westfield --Evelyn Beatrice Hall paraphrasing Voltaire | Chaos Berserker of Khorne
Re: config.h again
On Thu, 22 Mar 2001, James Henstridge wrote: On 21 Mar 2001, Lars Clausen wrote: Using this: grep -c 'config.h' `find . -name '*.c'` | grep ':0$' I found 67 places not including config.h (looks like Kjartan missed some in his patch to move the includes of config.h to the C files rather than headers). As for whether it should be '#include config.h' or '#include "config.h"', it looks like the autoconf manual recommends using config.h together with the -I. compiler flag (node "Configuration Headers" in the autoconf info documentation), so I suppose we should be using that. Ok, I'll fix that. Does that mean that all #includes should use ? I can fix that easily as well. -Lars -- Lars Clausen (http://shasta.cs.uiuc.edu/~lrclause) | Hrdgrim of Numenor "I do not agree with a word that you say, but I| Retainer of Sir Kegg will defend to the death your right to say it."| of Westfield --Evelyn Beatrice Hall paraphrasing Voltaire | Chaos Berserker of Khorne
Re: horizontal lines and EPS export and stuff :)
On Sun, 18 Mar 2001, Kasper B. Graversen wrote: Hi Dia-team! I just recently startet using Dia 0.86 (windows). A nice program, with some limitations though! I have two serious problems 1. How do I make vertical pointers from one box to another? It seems as if the arrow only want so point either up or down :-(( I'm hoping to improve the way zig-zag lines are initialized, but that requires some rework. In the meanwhile, here's a way around it (this will go into the FAQ): Draw the zig-zag line normally. At each end, use the middle mouse menu to add a segment. Drag the new segment handles towards the middle of the line. Remove the middle segment. -Lars -- Lars Clausen (http://shasta.cs.uiuc.edu/~lrclause) | Hrdgrim of Numenor "I do not agree with a word that you say, but I| Retainer of Sir Kegg will defend to the death your right to say it."| of Westfield --Evelyn Beatrice Hall paraphrasing Voltaire | Chaos Berserker of Khorne
Re: Dia Suggestion
On Tue, 20 Mar 2001, Mauricio de Castro wrote: Jeff Hornsberger wrote: It would really helpful for things like UML Use Case Diagrams if the arrows could be drawn diagonally. Thanks. -Jeff Would that be the lifeline line or the various aggregation/association etc lines? Changing the aggregation etc lines would be tricky, as they're defined as orthogonal lines. Changing the lifeline would be easier. -Lars -- Lars Clausen (http://shasta.cs.uiuc.edu/~lrclause) | Hrdgrim of Numenor "I do not agree with a word that you say, but I| Retainer of Sir Kegg will defend to the death your right to say it."| of Westfield --Evelyn Beatrice Hall paraphrasing Voltaire | Chaos Berserker of Khorne
Re: Compiling from CVS source
On Wed, 14 Mar 2001, Ben A. Hetland wrote: Hello, I'm trying to build the latest Dia from the CVS tree, but am unable to even generate the makefile needed. [...] Failing that, I read INSTALL that suggested I should run ./autogen.sh instead, which produced a whole lot of errors (variable not defined, undefined macros [AC_ stuff]), but basically ended up with the same two errors. It seems to me that it doesn't recognize the macros at all, but how is the script supposed to know where to look for them? I'm using RedHat 6.2 (fairly standard installation), and it seems like the required versions of the various tools mentioned in INSTALL (automake, autoconf, ...) are installed [rpm -q]. The errors from autogen.sh are the interesting ones, if you send them to the list, we should be able to figure it out. Have you checked that you have the othger required packages (libtool, for instance, seems to be less commonly installed). -Lars -- Lars Clausen (http://shasta.cs.uiuc.edu/~lrclause) | Hrdgrim of Numenor "I do not agree with a word that you say, but I| Retainer of Sir Kegg will defend to the death your right to say it."| of Westfield --Evelyn Beatrice Hall paraphrasing Voltaire | Chaos Berserker of Khorne
Re: Event trace (sequence) diagrams
On Mon, 5 Mar 2001, Richard Nelson wrote: Hi all! New to Dia and this forum. Apologise if this one has already been asked. I have been using Together Enterprise and Rational Rose on previous projects and I am evaluating Dia for a new project. I would like to know to create event trace diagrams in Dia. So far I have been using lifeline, object and message tools but I would have expected the property diagogue for the object instances to list classes that have been defined. I should also have expected the properties dialogue of the message instances to allow the selection of the operations available in the object instance the message is directed at (as defined by the class it is representing). Dia doesn't know that UML diagrams are anything else than funny figures with a specific name. However, these examples show how some more interaction can improve the UI. I suppose the Object and Message instance could look through the list of objects to find such information, but we probably need a new piece of infrastructure for that. All this ensures that the class names and message names used in event trace diagrams are valid. That would be a good thing, wouldn't it? It would require, however, that when a a class changes, all instances of it have incoming messages checked for validity. I hold that all this should be optional -- many people use UML in an entirely informal manner, and would only feel restricted by such syntax-checking. -Lars -- Lars Clausen (http://shasta.cs.uiuc.edu/~lrclause) | Hrdgrim of Numenor "I do not agree with a word that you say, but I| Retainer of Sir Kegg will defend to the death your right to say it."| of Westfield --Evelyn Beatrice Hall paraphrasing Voltaire | Chaos Berserker of Khorne
Re: Defaults dialogs
On Tue, 6 Mar 2001, James Henstridge wrote: On 5 Mar 2001, Lars Clausen wrote: I've looked a bit at making defaults dialogs from properties. I see three problems that need to be resolved: 1) Should the defaults be object-specific or name-based? I'd say name-based, but then we get the possibility of accidentally transferring defaults between sheets. Is that going to be a problem? 2) More serious: Each property should have a default value. If the defaults are name-based, there can only be one default value for any property of that name, which might not please everyone. In most cases, we probably want per property rather than per object (eg. at the moment, default colours, line styles, line widths and fonts are shared between all types). I am sure there are some cases where having things per object is useful as well. Maybe we could have it be the combination of the name and the default value that identifies a grouping? And maybe I should check how many cases there are of this before I do large nasty designs. 3) How does an object specify the default value? I'd like to add it to the PropDescription struct, but that would break binary compatability. Adding another struct for this would be plain silly. Well, at the moment PropDescription structs are static, whereas the default values for properties are mutable in dia. Is this feature worth preserving? How do we handle multiple objects types which should share a default, yet have separate (identical) PropDescription structs? What do you mean, the default values for properties are mutable? The only default values currently used (except for the standard properties) are hardcoded. With the glib 2.0 GObject properties, there is a g_param_value_set_default() function that can be used to modify the default for a GParamSpec, so this sort of thing will be possible when we move over to gtk 2.0 (and I think you can share a single GParamSpec between multiple object types on different branches of the type tree, which would handle the shared defaults problem). That does sound good. I'm looking forward to it. -Lars -- Lars Clausen (http://shasta.cs.uiuc.edu/~lrclause) | Hrdgrim of Numenor "I do not agree with a word that you say, but I| Retainer of Sir Kegg will defend to the death your right to say it."| of Westfield --Evelyn Beatrice Hall paraphrasing Voltaire | Chaos Berserker of Khorne
Re: Defaults dialogs
On Tue, 6 Mar 2001, James Henstridge wrote: On 5 Mar 2001, Lars Clausen wrote: On Tue, 6 Mar 2001, James Henstridge wrote: Maybe we could have it be the combination of the name and the default value that identifies a grouping? And maybe I should check how many cases there are of this before I do large nasty designs. I don't think it is safe to assume that properties of the same name on different object types are always going to be the same, and hence use the same default. However, in a lot of cases (the shared ones I mentioned above), it is. On the other hand, you would probably want all object types in one sheet share similarly-named properties. Maybe we could allow the definition of a 'defaults group' within which defaults are shared? Either per-object type or per-property. Or maybe just have each sheet share properties. Things like the "draw background" or "corner rounding" for the standard box have defaults that can be changed from the defaults dialog. Hence, they are mutable during the run of the program. All the other data in the PropDescription struct (name, description, prop type, etc) is fixed. I see what you mean. I don't see much reason to change this. -Lars -- Lars Clausen (http://shasta.cs.uiuc.edu/~lrclause) | Hrdgrim of Numenor "I do not agree with a word that you say, but I| Retainer of Sir Kegg will defend to the death your right to say it."| of Westfield --Evelyn Beatrice Hall paraphrasing Voltaire | Chaos Berserker of Khorne
Re: Postscript output size
On Tue, 6 Mar 2001, James Henstridge wrote: On 5 Mar 2001, Lars Clausen wrote: Well, I think that's overwhelmed by the 100 times as much data to send. Back in '90, printers were faster than workstations, and that was the time when PC's had self-extracting executables for the first time. So I think a simple RLE is worth it. I've lookad a bit at the code in XFig, it's taken from the old xwpick program. It features such niceties as a 4096x4096 maximum width and a 256 colors limit. The latter seems to be more difficult to get around, as the algorithm is deeply based on that. The version used in xfig (found in transfig) is slightly modified, but has the same limitations. An interesting project... I haven't checked what the algorithm in pnmtops is. It may not have the same limitation (or it may be the same postscript program). It is not really that high a priority for me at the moment. Ah, but you're not the one with a PS file that's 100 times too big. It takes about a minute just to generate it. pnmtops seems to have a more general algorithm, and is besides under GPL. I'll work off that. -Lars -- Lars Clausen (http://shasta.cs.uiuc.edu/~lrclause) | Hrdgrim of Numenor "I do not agree with a word that you say, but I| Retainer of Sir Kegg will defend to the death your right to say it."| of Westfield --Evelyn Beatrice Hall paraphrasing Voltaire | Chaos Berserker of Khorne
Re: Just a idea: connection lines insted of points
On Sun, 25 Feb 2001, Hans-Juergen Hay wrote: Hello, having used dia to play a little bit I had the idea that the concept of connection lines instead of points would bring many benefits. Instead of having a limited number of connection points at the border of objects connection lines would allow for unlimited different connections. Catching shouldn't be a bigger problem than with points. I could even imagine lines with different properties such as smoth lines, where connectors slide frictionless, sticky where connectors are fixed or inbetween where connectors slide just a little when the object is moved. Another enhancement might be generalising to polylines. This consept might be usefull in many kinds of diagramm. It's not a bad idea, but I'd probably prefer if the connectors autodistribute along a line, just to get it to look prettier. The standard line comes close to this, as you can add more connection points via the middle mouse menu. -Lars -- Lars Clausen (http://shasta.cs.uiuc.edu/~lrclause) | Hrdgrim of Numenor "I do not agree with a word that you say, but I| Retainer of Sir Kegg will defend to the death your right to say it."| of Westfield --Evelyn Beatrice Hall paraphrasing Voltaire | Chaos Berserker of Khorne
Visible bounding boxes!
As a debugging aid, I just put a bit into diagramdata.c that allows us to see the bounding boxes. It's toggled by a #ifdef in lib/diagramdata.c (ugly, I know, but I wanted it now:), and it shows some interesting behaviour with connection bounding boxes. It is of course turned off by default. I've also been working on making the UML objects more up-to-date wrt properties. Soon I'll add more color, I think:) -Lars -- Lars Clausen (http://shasta.cs.uiuc.edu/~lrclause) | Hrdgrim of Numenor "I do not agree with a word that you say, but I| Retainer of Sir Kegg will defend to the death your right to say it."| of Westfield --Evelyn Beatrice Hall paraphrasing Voltaire | Chaos Berserker of Khorne
Re: RENDER_BOUNDING_BOXES
On Fri, 23 Feb 2001, Cyrille Chepelov wrote: Lars, your code rocks. I'm really stupid I haven't thought of that before. Actually, the idea was my officemates, after we were looking at some funny behaviour on the Lifeline object. I'll have to fix at least arcs and bezier*; some objects (like lines) have overbig start_trans and end_trans when arrows are used (twice too big, precisely). Even a simple line (anything based on Connection, in fact) doesn't shrink its BB on the down and right sides. Oh, there it was. See if this change breaks the fine new extra_spacing code, please. Finally, there's the case of polylines and zigzags (same for beziers, I think), where the arrow head finds itself with a default orientation (0) when the last segment has a zero length (I'd vote for making the arrow disappear in that case, rather than extend the bounding box). Remember that as long as we can't query objects for orthogonality, the current arrow behaviour does allow one to get around the default orientations. I'm all for rewriting the arrow code, but that's really a thing for 1.1. -Lars -- Lars Clausen (http://shasta.cs.uiuc.edu/~lrclause) | Hrdgrim of Numenor "I do not agree with a word that you say, but I| Retainer of Sir Kegg will defend to the death your right to say it."| of Westfield --Evelyn Beatrice Hall paraphrasing Voltaire | Chaos Berserker of Khorne
More UML fixes
So I've fixed a couple more of the UML objects to use proper properties. I don't suppose the properties code supports something like what the Association object does, of grouping properties and placing them side by side? Not to mention what the Class object does... Also, the Branch object, which only has the ELEMENT_COMMON_PROPERTIES, shows up a property dialog with nothing in it, rather than 'This object has no properties'. UML users, please hit on the updated objects and see how they hold up. They should all do font stuff and bounding boxes nicely now (except for a funny late update in Component that I haven't figured out). -Lars -- Lars Clausen (http://shasta.cs.uiuc.edu/~lrclause) | Hrdgrim of Numenor "I do not agree with a word that you say, but I| Retainer of Sir Kegg will defend to the death your right to say it."| of Westfield --Evelyn Beatrice Hall paraphrasing Voltaire | Chaos Berserker of Khorne
Re: Bug in note tool
On 20 Feb 2001, Sebastian Fischmeister wrote: Hello. In the latest CVS version there is a "bug" in the note tool that messes up most of my drawings. I usually put only blanks in the note and then text on top of it, so I can change the font of the text. However, now a note filled with blanks does not grow larger (horizontally), although returns make it grow larger (vertically). If this is the UML note, then I just fixed it in CVS so that you can change font, font color and font size. (It was indeed almost there.) Then you shouldn't have to hack around it. -Lars -- Lars Clausen (http://shasta.cs.uiuc.edu/~lrclause) | Hrdgrim of Numenor "I do not agree with a word that you say, but I| Retainer of Sir Kegg will defend to the death your right to say it."| of Westfield --Evelyn Beatrice Hall paraphrasing Voltaire | Chaos Berserker of Khorne
Re: Bad exporting to EPS
On Tue, 20 Feb 2001, [EMAIL PROTECTED] wrote: Hi. My compliments to all who contribute to a very nice and useful program. I've just started using it and I have the problem that I cannot get the exported diagram recognized by gv ghostscript nor LyX (LateX in the end). I get the following ( using gv 3.5.8 ) Error: /undefined in 28,346000 Operand stack: Execution stack: %interp_exit .runexec2 --nostringval-- --nostringval-- --nostringval-- 2 %stopped_push --nostringval-- 2 3 %oparray_pop --nostringval-- --nostringval-- false 1 %stopped_push 1 3 %oparray_pop .runexec2 --nostringval-- --nostringval-- --nostringval-- 2 %stopped_push --nostringval-- --nostringval-- --nostringval-- Dictionary stack: --dict:904/941(G)-- --dict:0/20(G)-- --dict:84/200(L)-- Current allocation mode is local Current file position is 10761 Any suggestion will be greatly appreciated. Cheer from Italy... ^ Ah, there's you problem, right there. You're not living in the US:) Seriously, this is yet another case of overeager internationalization. PostScript assumes US style decimal separators, but by default we output internationalized separators. I'll fix that...there you go. It's now fixed in the CVS version. Other output formats may still have the same problem, we shall have to look at them in turn. -Lars -- Lars Clausen (http://shasta.cs.uiuc.edu/~lrclause) | Hrdgrim of Numenor "I do not agree with a word that you say, but I| Retainer of Sir Kegg will defend to the death your right to say it."| of Westfield --Evelyn Beatrice Hall paraphrasing Voltaire | Chaos Berserker of Khorne
Taking care of UML
Now I've gotten started on fixing properties for the UML objects, there are a couple of things that seem odd. For instance, both small_package and component has a stereotype and a name field that aren't used. Are these relics from a cut-and-paste, or are they supposed to be used somewhere? Alex and Hans seem to have worked on these, so can you enlighten me? -Lars -- Lars Clausen (http://shasta.cs.uiuc.edu/~lrclause) | Hrdgrim of Numenor "I do not agree with a word that you say, but I| Retainer of Sir Kegg will defend to the death your right to say it."| of Westfield --Evelyn Beatrice Hall paraphrasing Voltaire | Chaos Berserker of Khorne
Re: [Patch] UTF-8 EPS, and DPS (cont'd)
On Tue, 20 Feb 2001, Cyrille Chepelov wrote: Hi all, this patch is version 3 of my (DPS)/(EPS-UTF8) already too big patch. It does the following things : - (new) backs out Lars' setlocale() calls in render_eps.c - (new) use lazy colour changes in render_eps.c - build and switch encoding tables on the fly from UTF-8 strings in render_eps.[ch], with the required infrastructure. - a Display P*stscript renderer (optional). (Will use the UTF-8 infrastructure from render_eps.c RSN) So far (fifteen days), no loud shouts. I think I'll commit that soon. That's a challenge:) I took the patch to the source, and here's the result: In the patch, new files end up in the top-level directory. That's probably a CVS diff problem. I'm still trying to get the render_dps_wraps.h generated... that took some twiddling of the pswrap.mak file. Ok, there is goes. Loud shout: First thing I tried was importing a Fig file with various text in it. It turned out to use a font I don't have, which in turn crashed the DPS system, freezing Dia with it. This is the output: Unrecoverable error: invalidfont in findfont Operand stack: NimbusMonL-Regu Courier --nostringval-- Times-Roman 319469 Times-Roman Font Times-Roman --nostringval-- After that, I had to manually kill the DPS server to get Dia to start again. Now I'm trying with a blank canvas. It has the rightmost about 1/5 of the canvas black, and when I tried to scroll, DPS crashed again. And it's getting worse -- I'm guessing just killing the DPS servers isn't the happiest thing in the world:) But I can't find any docs on how to kill it properly. -Lars -- Lars Clausen (http://shasta.cs.uiuc.edu/~lrclause) | Hrdgrim of Numenor "I do not agree with a word that you say, but I| Retainer of Sir Kegg will defend to the death your right to say it."| of Westfield --Evelyn Beatrice Hall paraphrasing Voltaire | Chaos Berserker of Khorne
Re: What's the dia web site address?
On Mon, 19 Feb 2001, Jacek Pliszka wrote: On Mon, 19 Feb 2001, Scott Vetter wrote: What is the Dia web site address? Go to www.google.com enter: dia Or even: dia program and click "I'm feeling lucky". Google rules! -Lars -- Lars Clausen (http://shasta.cs.uiuc.edu/~lrclause) | Hrdgrim of Numenor "I do not agree with a word that you say, but I| Retainer of Sir Kegg will defend to the death your right to say it."| of Westfield --Evelyn Beatrice Hall paraphrasing Voltaire | Chaos Berserker of Khorne
Re: XFig import plug-in
On 19 Feb 2001, Lars Clausen wrote: I've now commited the first rough version of the XFig import plugin. Several things aren't implemented yet (such as anything poly-like), but quite a few things work. The boxes seem to have trouble with overly rounded corners, and most wierdly, the text objects don't get any text in them. All the other text properties (font, color etc) are set nicely, but the text never appears. I'm setting the text with props[0].name = "text"; props[0].type = PROP_TYPE_STRING; PROP_VALUE_STRING(props[0]) = strdup(text); new_obj-ops-set_props(new_obj, props, 1); -- is there something I've missed there? (Apart from probably not needing the strdup.) Ok, I found the bug. It was in textobj.c, it didn't do the quarks right, so it missed the text property for set_props. Fixed now, and I'm rapidly approaching a useable state for the imports. -Lars -- Lars Clausen (http://shasta.cs.uiuc.edu/~lrclause) | Hrdgrim of Numenor "I do not agree with a word that you say, but I| Retainer of Sir Kegg will defend to the death your right to say it."| of Westfield --Evelyn Beatrice Hall paraphrasing Voltaire | Chaos Berserker of Khorne
Creating objects for import
I'm working on XFig import (finally!), and while the simple objects like boxes and ellipses are easy enough, the polygon is not. If I were to only use the external interface to Standard - Polygon, I can't add extra corners. If I also use the functions from the middle mouse menu, I can add corners, but with little control over how the corners are ordered (since it picks the closest segment always). So I'm this close to adding a new standard function, basic_create(Point[], int), that will create the object directly rather than using handles. The function takes an array of points that the object is created from, and the number of points in the array. Exactly how the points are use is defined by the object. Anyone have problems with this? -Lars -- Lars Clausen (http://shasta.cs.uiuc.edu/~lrclause) | Hrdgrim of Numenor "I do not agree with a word that you say, but I| Retainer of Sir Kegg will defend to the death your right to say it."| of Westfield --Evelyn Beatrice Hall paraphrasing Voltaire | Chaos Berserker of Khorne
Re: [wishlist] moving/resizing/drawing only across one axis
On Fri, 9 Feb 2001, Vlad Harchev wrote: Hi, Most vector-drawing programs (and Visio2k too) allow to move things/resize in one direction/draw lines that are either horizontal or vertical if some keyboard modifier is pressed (I don't remmeber which one is the most used one for this - may be shift?). I find such ability very useful. Is such feature planned for next releases of Dia? It seems to be easily implementable by persons who have knowledge of dia's internals.. 'Tis already available, with Ctrl-mouse. Enjoy. -Lars -- Lars Clausen (http://shasta.cs.uiuc.edu/~lrclause) | Hrdgrim of Numenor "I do not agree with a word that you say, but I| Retainer of Sir Kegg will defend to the death your right to say it."| of Westfield --Evelyn Beatrice Hall paraphrasing Voltaire | Chaos Berserker of Khorne
Re: moving/resizing/drawing only across one axis
On Fri, 09 Feb 2001, Hubert Figuiere wrote: Lars Clausen crit: Most vector-drawing programs (and Visio2k too) allow to move things/resize in one direction/draw lines that are either horizontal or vertical if some keyboard modifier is pressed (I don't remmeber which one is the most used one for this - may be shift?). I find such ability very useful. Is such feature planned for next releases of Dia? It seems to be easily implementable by persons who have knowledge of dia's internals.. 'Tis already available, with Ctrl-mouse. Enjoy. The problem is that it does not work like it should (IMHO) when modifing a shape. Constraint should not be on cursor move but on the shape. For example, ctrl-move to a line anchor should make a line within PI / 4 angle. Ctrl-resize should resize proportionnaly. I see. That would require quite a bit more recoding, as we'd have to have the objects decide what Ctrl-something means for them. Just my 2 Euros. Hub It's getting expensive to express your opinion in Europe, isn't it?:) -Lars -- Lars Clausen (http://shasta.cs.uiuc.edu/~lrclause) | Hrdgrim of Numenor "I do not agree with a word that you say, but I| Retainer of Sir Kegg will defend to the death your right to say it."| of Westfield --Evelyn Beatrice Hall paraphrasing Voltaire | Chaos Berserker of Khorne
Re: hub object modification
On Wed, 07 Feb 2001, Hubert Figuiere wrote: Hello Here is a patch against 0.86 version of Dia that provide 8 handles to the "Network - Hub" object corresponding to the 8 ports drawn. This make it a little bit more useful Q: how is the compatibility with older versions ? Hub Patch applied and committed. Compatibility is good -- old diagrams load, new diagrams load in old versions, but it complains about connection points. If you make a diagram in the new version, then save it in the old version, you lose the connections, but that's to be expected. BTW, I notice that Gimp has a repetition count on its error dialogs. This may be a thing to consider, to avoid getting a gazillion dialogs for some small compatibility change. -Lars -- Lars Clausen (http://shasta.cs.uiuc.edu/~lrclause) | Hrdgrim of Numenor "I do not agree with a word that you say, but I| Retainer of Sir Kegg will defend to the death your right to say it."| of Westfield --Evelyn Beatrice Hall paraphrasing Voltaire | Chaos Berserker of Khorne
Re: Invisible fonts
On Wed, 7 Feb 2001, Jan Ploski wrote: Another thing that caught my attention is that the "object properties" dialog doesn't disappear when I hit Enter. I'd expect it to accept settings and let it go, but I have to click on "Apply" explicitly. (Perhaps it's just me, if not, that's another request for enhancement.) No, that's just a silly behaviour, IMHO. Anytime a property has been changed, any 'enter' press should go to either Apply or OK (not quite sure which -- probably OK). Currently, some of the properties eat the enter press. -Lars -- Lars Clausen (http://shasta.cs.uiuc.edu/~lrclause) | Hrdgrim of Numenor "I do not agree with a word that you say, but I| Retainer of Sir Kegg will defend to the death your right to say it."| of Westfield --Evelyn Beatrice Hall paraphrasing Voltaire | Chaos Berserker of Khorne
Re: FAQ addition
On Tue, 06 Feb 2001, Hubert Figuiere wrote: Here is FAQ that it would be nice to see added to the Dia FAQ: Q: Can Dia open Visio .vsd files ? A: No, it can't. Visio file format is a completely prioprietary and undocumented file format. So it is really difficult to write code to read it. By seeing the bounty available at http://visar.csustan.edu/bazaar/bazaar_dateoffers.html for a reverse engineering of the file, we suspect it is not an easy task. We really would like to be able to do so. If someone can add it ? Hub Thank you, it has been added. The site should be updated at midnight CET. -Lars -- Lars Clausen (http://shasta.cs.uiuc.edu/~lrclause) | Hrdgrim of Numenor "I do not agree with a word that you say, but I| Retainer of Sir Kegg will defend to the death your right to say it."| of Westfield --Evelyn Beatrice Hall paraphrasing Voltaire | Chaos Berserker of Khorne
Re: Visio vsd files into dia?
On Mon, 05 Feb 2001, Emmett Lazich wrote: Can anyone offer a suggestion for importing Visio vsd files into dia? I thought I had it licked when I noticed that Visio appeared to be able to export in postscript format(s) and pstoedit can convert such postscript files into dxf format. However vector file format exporting in Visio 5.0 seems to be broken. Besides, it would be more convenient to be able to do this vsd to dia import without requiring Visio at all! We would add vsd import in an eyeblink except that the format is proprietary just like most other Microsoft programs formats. Anyone who can figure out the format will be awarded the Dia Medal of Honor. -Lars -- Lars Clausen (http://shasta.cs.uiuc.edu/~lrclause) | Hrdgrim of Numenor "I do not agree with a word that you say, but I| Retainer of Sir Kegg will defend to the death your right to say it."| of Westfield --Evelyn Beatrice Hall paraphrasing Voltaire | Chaos Berserker of Khorne
Re: Visio vsd files into dia?
On Mon, 5 Feb 2001, Hubert Figuiere wrote: On Mon, 05 Feb 2001, Emmett Lazich wrote: Can anyone offer a suggestion for importing Visio vsd files into dia? I thought I had it licked when I noticed that Visio appeared to be able to export in postscript format(s) and pstoedit can convert such postscript files into dxf format. However vector file format exporting in Visio 5.0 seems to be broken. Besides, it would be more convenient to be able to do this vsd to dia import without requiring Visio at all! We would add vsd import in an eyeblink except that the format is proprietary just like most other Microsoft programs formats. Not that I want to defend Microsoft, but Visio file format was completely undocumented even before Microsoft bought the product (and its company). I believe that it true. It's not likely to become any more open now, though. Even worse, Visio was really hypocrite by creating a task force to reverse engeenier the AutoCAD binary file format... And speaking Microsoft, they provide (perhaps incomplete and vague) documentation of several of their file format, which we can't tell from some other companies. I tried looking for the Word format. The documentation isn't even available for subscribers to MSDN. Emmett, I would encourage you to look at the archive list this as already be discussed. And I believe there's still a $2000 reward for reverse engineering it. -Lars -- Lars Clausen (http://shasta.cs.uiuc.edu/~lrclause) | Hrdgrim of Numenor "I do not agree with a word that you say, but I| Retainer of Sir Kegg will defend to the death your right to say it."| of Westfield --Evelyn Beatrice Hall paraphrasing Voltaire | Chaos Berserker of Khorne
Re: Windows build
On Wed, 31 Jan 2001, Jim Thompson wrote: Is there a Windows build of dia available for download? Yes, on URL:http://hans.breuer.org/dia/. -Lars -- Lars Clausen (http://shasta.cs.uiuc.edu/~lrclause) | Hrdgrim of Numenor "I do not agree with a word that you say, but I| Retainer of Sir Kegg will defend to the death your right to say it."| of Westfield --Evelyn Beatrice Hall paraphrasing Voltaire | Chaos Berserker of Khorne
Re: Background Pattern
On Thu, 1 Feb 2001, Hilmar Strickfaden wrote: Hello everybody First of all I want to thank you for the hint on the Venn-Diagrams. Now I have a new question. Is it possible to use a background patterns for the objects instead of simple colours ? Not yet, though there's nothing to prevent it from happening. -Lars -- Lars Clausen (http://shasta.cs.uiuc.edu/~lrclause) | Hrdgrim of Numenor "I do not agree with a word that you say, but I| Retainer of Sir Kegg will defend to the death your right to say it."| of Westfield --Evelyn Beatrice Hall paraphrasing Voltaire | Chaos Berserker of Khorne
Re: Patch for copy-and-paste of ellipse and zigzagline
On Wed, 31 Jan 2001, Martin Hawlisch wrote: Hi! Copy and paste is not working properly for zigzagline and ellipse as the dash length is not copied. As I don't have CVS-write access the patch is appended: Thanks, it has been applied now. -Lars -- Lars Clausen (http://shasta.cs.uiuc.edu/~lrclause) | Hrdgrim of Numenor "I do not agree with a word that you say, but I| Retainer of Sir Kegg will defend to the death your right to say it."| of Westfield --Evelyn Beatrice Hall paraphrasing Voltaire | Chaos Berserker of Khorne
Re: dia is great but ...
On Mon, 29 Jan 2001, Hubert Figuiere wrote: Which reminds me: When you have made changes and subsequently undone them, Dia still thinks the diagram is modified. Shouldn't diagram_modified_exists consider a diagram unmodified if it has no undo history? Yes. Better: why wouldn't we alow undo past saving ? Emacs does this and sometimes this helps. This involve to have a save marker in the undo stack and not flushing the undo stack at save. We don't flush the undo stack at save. But a save marker in the undo stack would mean that we wouldn't need the diagram-modified field anymore. -Lars -- Lars Clausen (http://shasta.cs.uiuc.edu/~lrclause) | Hrdgrim of Numenor "I do not agree with a word that you say, but I| Retainer of Sir Kegg will defend to the death your right to say it."| of Westfield --Evelyn Beatrice Hall paraphrasing Voltaire | Chaos Berserker of Khorne
Re: dia is great but ...
On Thu, 25 Jan 2001, Patrick Amirian wrote: Hi, I'm using Dia and I think it's a great tool, I enjoy using it and it's simple. The problem I'm having is to find a way to publish my network graphs on our ethernet and... I can't. how can I save my graph into a png or jpg file ? can it be done ? Yes, select File/Export on the image menu. Then pick your file format of preference. I think I'm just quietly gonna combine Save As and Export to avoid these problems in the future. -Lars -- Lars Clausen (http://shasta.cs.uiuc.edu/~lrclause) | Hrdgrim of Numenor "I do not agree with a word that you say, but I| Retainer of Sir Kegg will defend to the death your right to say it."| of Westfield --Evelyn Beatrice Hall paraphrasing Voltaire | Chaos Berserker of Khorne