This is a rough record of the main subjects of discussion. Copied with slight changes from http://piratepad.net/inkscape The agenda turned out to be too long for a single meeting, and several points were left untouched.
1. Inkscape 0.48.3.1 packaging Need people (2 or 3) to create an installer on win32 and OSX. How about UweSch ? Tweenk is making a package on a VM (for win). Quite slow need 1h. Discussion about tool (btool, MSYS..) "we have 2 ways forward, fix our autotools scripts to work with MSYS (should be doable since they have recently updated their Perl port) or move to a different build system" Resume: ScislaC and Tweenk will see about building the Win32 packages. They need to figure out if the project could invest in virtualized OS copies so they can knock specific builds out (like OSX). If Michael W. doesn't have time/resources, su_v could provide 32bit Intel (Leopard) and 64bit Intel (Lion) packages (no Universal packages) Unit testing: we need to get our unit tests cleaned up and all passing. Then a continuous build system can take care of that whole one-os-breaks-the-others issue. Breaking OS often stems from #ifdef or package issues, so is not solved by unit testing. Unit testing + continuous build systems on multiple OS's catches #ifdef breakage, etc.
2. Inkscape 0.49 2.1 - Remaining refactoring goals Cairo and render caching : still some bugs to fix Gradient rendering: kuribas offers to chip in to help fix the banding. Idea is to add it as an option in exporting. Needs to be implemented in Cairo. disadvantages: speed, filesize, consistency between outputs, shouldn't be done partially (only on gradients f.e.) proposal (Tweenk): implement a GEGL renderer as an alternative to the Cairo renderer for use in export (running 16 bit engine under the hood)
2.1.1 - Handling of svg:switch not as a kind of group, but as a "transparent" element that is effectively replaced by the best choice in the SP tree (ScislaC): we need to be a solid test ground for SVG2 ... switch helps make it happen. (Tweenk): there is an asymmetry between the "base path" and "path parameters": the "base path" is stored in the svg:path that represents the result, while all other path parameters are stored in, IIRC, inkscape:liveeffect (Tweenk): the primary use case for switch in Inkscape would be to put 2 elements in it where the first is an inkscape-specific XML element while the second is an SVG fallback browser implementation is not at par yet - this isn't usable on the web for now. Some playing/testing with Firefox: (Tweenk): looks like FF works as I intended only if I put the SVG element first, and the Inkscape-specific XML second link: https://developer.mozilla.org/en/SVG/Element/switch (Tweenk): if I put Inkscape XML in a foreignObject it works properly (MenTaLguY): foreignObject is probably the technically correct thing anyway (Tweenk): there would be two elements in the <switch>, a inkscape:something inside <foreignObject> that would contain flowframe info and a <g> that would contain SVG 1.1 text (...and ScislaC nods)
2.1.2 - SVG 1.1 fallback for flowtext (probably depends on the above) see above - 2.1.1
2.1.3 - UI & Document coordinates the problem in short: where do you set the (0,0) point and what direction? Currently: (Inkscape standard): bottom left corner = (0,0) and up is positive versus (SVG standard): top left corner = (0,0) and down is positive. Current fix breaks a few things stored in desktop coordinates: grids, 3D boxes and guides, rows and columns, view coordinates, ... (JonCruz): "I'd say we need to have a wiki page to start listing what can be affected, and the use cases we care about" Changes shouldn't alter the SVG, but be purely interface (preferrably the document origin should be configurable on a per document basis) 3D box: fix or remove and then rewrite, if a new one is written, keep in mind http://dev.w3.org/csswg/css3-transforms/ wiki page to summarize broken areas, general use cases, etc. for guiding the coord fixes: http://wiki.inkscape.org/wiki/index.php/UI_and_Document_coordinates 2.1.4 - Markers matching stroke by default (user preference vs current default behavior) See SVG2 below. 2.1.5 - Better architecture for Live Path Eeffects (LPE; similar to SVG 2.0 vector effects) 2.1.6 - User visible version numbering suggestion of switching to a different scheme after 0.49 is released 1.0 conditions: fix most annoying problems: flowtext, coords, marker coloring discussion background: http://thread.gmane.org/gmane.comp.graphics.inkscape.devel/31650 and take 2: http://thread.gmane.org/gmane.comp.graphics.inkscape.devel/36540 2.2 - Status 0.49 should contain the coords fix (but it should be forward compatible with origin being placed in another spot) Flowtext fallback is too complicated and there's already too much new stuff in 0.49 Lack of mipmapping / downscaling in Cairo: critical bug WIKI page is needed to work on this: http://wiki.inkscape.org/wiki/index.php/Mipmapping 2.3 - Release Plan 3. - GSOC 2012 3.1 - Focus / Goals Pick coders who have better chances of sticking around and delivering instead of those who promise super-features Focus on completing all selected projects successfully with benefit to the project rather than on adding flashy new stuff Preference on structural type projects over features 3.3 - Mentoring Available mentors: Johan E. (dutch, english, german), Tweenk (english, polish) (can also be a student), Tav (English, W. African Pidgin), JonCruz (english) Available mentors should be pinged: JazzyNico, pygmee, steren. i guess we can manage some "french team" Not available: Jasper vd G. Available as admin: simmie 3.4 - Getting new students JonCruz mentioned something about African students - African countries have a lot of open source activity, but most of it is in French-speaking countries (language barrier) -> JazzyNico, pygmee and (Steren -if available) can speak French : need to find available african too but we can have help from some organization for this :) For GSoC purposes, 2Geom should be part of Inkscape - interesting potential projects with concrete deliverables; relates to point 8 (to be discussed in the future) Strictly enforce two patches rule Have IRC meetings with mentors/students as part of our process (Johan: I meant this as part of our *selection* process)
4. SVG2 and CSS3 - not discussed thoroughly due to lack of time. 4.1 - Implementation strategies 4.1 .1 - Fallback support for older versions of the spec and inkscape 4.1 .2 - Specific features (markers inheriting properties for example) 4.2 - Low hanging fruit Marker color matching stroke. Named colors. CSS3 2d transforms (duplicates current attributes) 4.3 - Most important features to our userbase 4.3.1 - Helping Tavmjong with the Mesh branch 4.4 - Initial Vector Effects support (non-destructive boolean ops might be a good target) - relates to Live Path Effects 4.5 Pushing for wanted features in SVG2. - multipage support
8. What we need in upstream libraries (including 2geom) 2geom: coders better documentation of mathematics behind path operations more unit tests port existing tests to Google Test Framework? provide skeleton test example libgdl: named icons in dock item labels there's a bug in the code that we currently have imported into our codebase that makes 64 bit architectures crash: https://bugs.launchpad.net/inkscape/+bug/933058/comments/10 cairo: bitmap downscaling / mipmapping libcroco: ability to work with other XML tree implementations this is provided by our in-tree cr-sel-eng (selection engine) modification
11. Platform specific issues (and responsibilities) - see point 1 12. Regular IRC Meetings (monthly is ideal, quarterly would be as a good minimum) general consensus on 1 or 2 times per month. suggestion to let timeslots alternate (f.e. 8AM and 8PM UTC) to accomodate ppl worldwide
Various points: * do we need more focus on 'polish' instead of new features? see Bruce Perens' keynote on Linux.conf.au http://www.youtube.com/watch?v=Uoum-DHO7S8 * long-winded discussion about preferences system * this discussion then spilled over into the embed/link debate
Participants (list might be incomplete): jazzynico, Johan Engelen (johan_e), Jon A. Cruz, Jurgentje, Kris, Nikita Kitaev, parasidasaeidae, pygmee, ScislaC, simarilius, su_v, Tavmjong, Tweenk, yemanjalisa