December 2003 - Inkscape Devel

MinGW build info
by Bob Jamison 19 Dec '03

19 Dec '03

Hi guys, I have a clean MinGW build working. Here is the result: http://troi.lincom-asg.com/~rjamison/files/inkscape.zip It is built with last night's CVS, and should include all of the necessary dependencies. It would be good for someone to check that out. Here is the current set of Gtk libs necessary to link: http://troi.lincom-asg.com/~rjamison/files/gtk2.zip Also, here are the Makfiles used to build it: http://troi.lincom-asg.com/~rjamison/files/makefiles.zip Here are the diffs of the tweaks necessary to get it to compile: http://troi.lincom-asg.com/~rjamison/files/win32.diff And finally, here are the codepages which should be unpacked into the /src/libnrtype lib: http://troi.lincom-asg.com/~rjamison/files/nrtype-codepages.zip Although any person can take these and build on Win32 with MinGW, IMHO, it would be very beneficial to integrate these into the tree. Far better than having a "Win32 Port," would be to have a "Win32 Build" implying that the source tree is always buildable on Win32. Hope this is useful to someone. Bob Jamison Titan Corp.

1 0

Re: [Inkscape-devel] spiral???
by bulia byak 19 Dec '03

19 Dec '03

> > I must say that arrow keys moving is already much more powerful in >Inkscape > > (both by pixels and by fixed increments), while using arrow keys for > > rotating objects is a very stupid idea, imho. Arrow keys just don't >stand > > for rotation. I'm now implementing a comprehensive set of shortcuts for > > rotation using [ and ] keys with various modifiers. > >[clearly i need to build and use InkScape a whole lot more and not be so >badly uninformed about how things like the spiral tool work, but the >machine i use for most of the day and write my email on has a fairly >minimal selection of software] With regard to shortcuts, you can consult http://inkscape.org/cgi-bin/wiki.pl?KeyboardShortcuts which I keep up-to-date as I add more shortcuts. >I'd need to think about it some more but I don't think using the arrow keys >+ modifier for rotation is such a terrible idea but I'll wait and see (I >may well be wrong). There are 2 rotation directions, but 4 arrow keys. Confusing. >I think there is plenty of room to use those >modifiers to overload the arrow keys to allow rotation. In fact, almost all possible modifiers are already used - see the wiki page. Ctrl-arrows pan the canvas; shift, alt, and shift-alt move objects. >However if you want to do things with [ and ] you will definately need to >at least check the AZERTY keyboard layout and perhaps others to make sure >that it is not too awkward to use those keys. I think no matter where these keys are located (provided they are present at all), they are logical enough for this function. >Even if you do decide to use different keys it would certainly be nice to >be able to move things by 1 pixels and N pixels increments (the N pixels >is configurable not just fixed to 10 which i realised after writing the >aformentioned notes). The 10 multiplier may be made configurable, but I don't see much sense in that. It's like coins: you can pay any amount if you have 10-cent and 1-cent coins. >Thanks for the reply, I am very glad that you are also interested in >providing users with keyboard accelerators for most functionality >and I think it is a very good idea to have default keybinding for almost >everything. Yes, that seems to be my speciality in Inkscape :) _________________________________________________________________ STOP MORE SPAM with the new MSN 8 and get 2 months FREE* http://join.msn.com/?page=dept/bcomm&pgmarket=en-ca&RU=http%3a%2f%2fjoin.ms…

3 2

RE: [Inkscape-devel] shift+click to select broken
by bulia byak 19 Dec '03

19 Dec '03

>>Just noticed that shift+click to select multiple objects is broken in >>HEAD... any suggestions for a fix? > >Oops. My fault. Will fix now. Fixed, please test. I had to make use of sp_document_partial_items_in_box() for this, which was in the code but never used, and fixed a bug in it so it could actually work. While at it, I changed the behavior of shift-rubberband. Previously it toggled the selected status of all objects it affected, now it always adds to selection. Rationale: the semantics of rubberband ("include everything within this box") and toggle ("deselect what was selected and vice versa") do not mix, imho. Especially given simple rubberband (without shift) always selects objects, regardless of their past status. Shift-click, however, still toggles selection status, so you can select many objects with (shift-)rubberband and then deselect individual ones with shift-clicks. Comments? And the main change which was the reason to dig into this code at all: now if Shift is pressed before you start dragging, you always get rubberband, even if you start from an object. I admit I stole this from Xara, but it's VERY convenient. Without it, if you had a big background object in your drawing, you simply could not use rubberband because you always ended up dragging the background. Now you can, just press Esc and then Shift-rubberband for a new selection. _________________________________________________________________ Add photos to your e-mail with MSN 8. Get 2 months FREE*. http://join.msn.com/?page=features/photos&pgmarket=en-ca&RU=http%3a%2f%2fjo…

3 3

Re: [Inkscape-devel] questions
by bulia byak 19 Dec '03

19 Dec '03

> > 2. I want to enable ctrl-click to select within group, but I don't see >how > > it can be done. I've delved deep into the arena code and found that if a > > group is "transparent" then its members can be selected one by one. But >how > > can I do that on ctrl-click? In select-context.c, the handler that >checks > > what button/key was pressed seems to already get a selected item as an > > argument from the arena code. Will I have to analyze this preselected >item > > on my own, or is there a more elegant way? > >This is something I'd like to work on personally, as part of the >continuing layers work. I'm still trying to determine what a "clean" >solution would be, though. Actually, analyzing the object under mouseclick and finding out what part of selection was clicked is pretty easy using document.cpp functions, without any help from arena code. I think I could write this feature now, but if you don't want me to mess with this, I'm fine :) >This sounds like a compiler bug. What version of g++ are you using? >I've been unable to replicate the problem using g++ 3.3.2. Mine is 3.3.1. Do you think it's worth the trouble to protect 3.3.1 users from this crash (which is admittedly rare, you'll need to scale a text string to microscopic sizes to get it)? _________________________________________________________________ Help STOP SPAM with the new MSN 8 and get 2 months FREE* http://join.msn.com/?page=dept/bcomm&pgmarket=en-ca&RU=http%3a%2f%2fjoin.ms…

2 1

questions
by bulia byak 19 Dec '03

19 Dec '03

With our discussions being so lively :) perhaps something gets lost. A couple of my recent questions were not answered, so I'm not reposting them in the hope that they reach knowledgeable people who might have missed them the first time. 1. Regarding this bug: http://sourceforge.net/tracker/index.php?func=detail&aid=860202&group_id=93… does anyone else think that document.cpp must store the absolute path already, i.e. instead of document->uri = g_strdup (uri); in sp_document_create we should have something like document->uri = abs_path (uri, base); That would fix the above bug. If so what standard function is there for creating an absolute path? I could not find one in glib. 2. I want to enable ctrl-click to select within group, but I don't see how it can be done. I've delved deep into the arena code and found that if a group is "transparent" then its members can be selected one by one. But how can I do that on ctrl-click? In select-context.c, the handler that checks what button/key was pressed seems to already get a selected item as an argument from the arena code. Will I have to analyze this preselected item on my own, or is there a more elegant way? 3. "Black magic": has anyone managed to reproduce this bug: https://sourceforge.net/tracker/index.php?func=detail&aid=859364&group_id=9… and its "fix": add an empty g_print("") to line 204 of libnrtype/nr-rasterfont.c: NRRectF * nr_rasterfont_generic_glyph_area_get (NRRasterFont *rf, unsigned int glyph, NRRectF *area) { NRRFGlyphSlot *slot; glyph = CLAMP (glyph, 0, rf->nglyphs); g_print(""); slot = nr_rasterfont_ensure_glyph_slot (rf, glyph, NR_RASTERFONT_BBOX_FLAG | NR_RASTERFONT_GMAP_FLAG); If I don't hear from anyone I'm going to commit this fix anyway (with appropriate comments), because, at least, it will not break anything. A better solution would of course be welcome to replace that temporary fix. _________________________________________________________________ MSN 8 with e-mail virus protection service: 2 months FREE* http://join.msn.com/?page=features/virus&pgmarket=en-ca&RU=http%3a%2f%2fjoi…

2 1

Re: [Inkscape-devel] spiral???
by bulia byak 19 Dec '03

19 Dec '03

> > the star, and maybe a polygon tool of some sort for n-th sided polygons. > >I've been meaning to suggest that it would be useful to have a Polygon as >well as a "Star" Tool (pentagram, poly* => Polygram?). >It brings the feature closer to the user, easier to discover and use even >and worth the slight repitition. You can create a polygon with the star tool now, if you use node edit to move the two control points together. However it's a good idea to provide a shortcut for that action, I will add it to TODO. >If the current spiral tool does not do enough to be interesting to you and >you want ways to make it more poweful/usefull I am linking to an example >file (svg and png) and a screenshot of the Spiral Tool from Jasc Web Draw >which has lots of options (curved or flat sides, number of turns, number >of rotations, linear or logarithmic expansion etc). All of these options, with the exception of flat-sided spirals, are possible with Inkscape. It's even more flexible: instead of choosing between linear and logarithmic only, it allows you to set the expansion value precisely (both above and below 1, with 1 being linear), and you can vary the number of rotations interactively on the canvas (switch to node edit). Overall, it's a very good tool. Although the flat-sided spirals would be nice to have too. >Although I keep mentioning Jasc Web Draw I have to say that it is a flawed >program missing all kinds of details but I do think it contains a few >fresh ideas and I will try and take more screenshots. I probably >shouldn't speculate like this but I'd really like to see a stable Inkscape >1.0.0 [1] declared this side of six months and give them a run for their >money! here are those links. I found this on your site: http://www.maths.tcd.ie/~horkana/inkscape/jasc-webdraw-tooltips.txt I must say that arrow keys moving is already much more powerful in Inkscape (both by pixels and by fixed increments), while using arrow keys for rotating objects is a very stupid idea, imho. Arrow keys just don't stand for rotation. I'm now implementing a comprehensive set of shortcuts for rotation using [ and ] keys with various modifiers. _________________________________________________________________ Add photos to your e-mail with MSN 8. Get 2 months FREE*. http://join.msn.com/?page=features/photos&pgmarket=en-ca&RU=http%3a%2f%2fjo…

4 3

verb cleanup
by MenTaLguY 19 Dec '03

19 Dec '03

Just a quick note to let y'all know that I've finished the work of making SPActions per-view. Now, if you're implementing a new verb, please use SP_DESKTOP (sp_action_get_verb(action)) to get the associated desktop, instead of using SP_ACTIVE_DESKTOP. -mental

1 0

RE: [Inkscape-devel] shift+click to select broken
by bulia byak 19 Dec '03

19 Dec '03

>Just noticed that shift+click to select multiple objects is broken in >HEAD... any suggestions for a fix? Oops. My fault. Will fix now. _________________________________________________________________ Add photos to your messages with MSN 8. Get 2 months FREE*. http://join.msn.com/?page=dept/features&pgmarket=en-ca&RU=http%3a%2f%2fjoin…

1 0

shift+click to select broken
by MenTaLguY 19 Dec '03

19 Dec '03

Just noticed that shift+click to select multiple objects is broken in HEAD... any suggestions for a fix? -mental

1 0

Re: [Sodipodi-list] Re: 0.8 multiplier
by MenTaLguY 19 Dec '03

19 Dec '03

On Sat, 2003-12-13 at 05:10, Lauris Kaplinski wrote: > > (this is why all examples in the SVG spec specify a viewBox) > > > > Thinking about this, if Sodipodi intends to be completely interoperable, > > it needs to always write a default viewBox according to the base > > units/points mapping it imposes. > > > > e.g. for a document 100ptx100pt, the viewBox should be "0 0 125 125". > > Are you saying, that 'width' and 'height' attributes of outermost > SVG have different semantics, as the same attributes in inner elements? > AFAIK it is not the case, so viewBoxonly changes the meaning of: > A) Percentages (as these are not defined as % of viewbox) > B) Base units by ratio of width/viewBox_width and height/viewBox_height > - but as width and height are in relative units, this is no > different that applying 'transform'. Consider these two SVG documents (I'll leave out namespace/XML declarations and other boilerplate for brevity): <svg width="100pt" height="100pt"> <rect x="12.5" y="12.5" width="100" height="100" fill="black" stroke="none"/> <rect x="10pt" y="10pt" width="80pt" height="80pt" fill="none" stroke="red" stroke-width="1pt"/> </svg> and: <svg width="100pt" height="100pt" viewBox="0 0 125 125"> <rect x="12.5" y="12.5" width="100" height="100" fill="black" stroke="none"/> <rect x="10pt" y="10pt" width="80pt" height="80pt" fill="none" stroke="red" stroke-width="1pt"/> </svg> Well, hrm. Having just tried this with Batik, I think it reveals a fundamental misunderstanding I had about viewBox. So, for the document root it appears that the relationship between pixel and non-pixel measurements can never be guaranteed to be fixed between different UAs/display devices. Doesn't this mean that Sodipodi should probably be using pixel measurements, not real-world units, for the height/width properties of the root SVG element? (and probably using e.g. sodipodi:height and sodipodi:width, or possibly sodipodi:dpi to derive the real-world measurements the UI uses?) Otherwise, in other SVG UAs, the relationship between the drawing and page size is not guaranteed to be the same. -mental

1 0