thanks for planning this thoroughly! Answers below.
Am 12.08.2019 um 21:13 schrieb Marc Jeanmougin:
== Extensions (Martin, Patrick, others) ==
What is the state of extensions (#156) ? In particular, their translatability (#333) ? Is the current submodule commit in a "good state" ? What to do with the old uniconvertor extensions, do we update them to uniconvertor 2.0rc5 ?
While Martin and others did (and are still doing) a great job on the bundled Python extensions I'm currently working on the extension implementation in Inkscape itself.
I just finished one big part (.inx format updates and the internal logic in Inkscape to parse preferences/parameters form it while also creating nice GUIs) which is ready for review at https://gitlab.com/inkscape/inkscape/merge_requests/808
As Martin already mentioned translations are basically handled as before (still use inkscapes .pot/.po files), only the way we mark strings for translation in the.inx changed. Once the MR is merged it will be a trivial job to update the .inx files with some search&replace, although the code should be fully backwards-compatible to earlier .inx files.
Regarding splitting the translations of extensions into the extension repo (and also allowing third-party extensions to ship their own translations): The current MR implements the syntax, but the actual functionality is not included yet. However I realized while working on it, that I already have done most of the required refactoring, minimizing the effort that remains for finishing the feature. If we feel like this is a welcome feature for 1.0 already, we can probably implement it in time instead of delaying it until 1.1 as I initially planned. I'm starting to think that this would actually be nice, as we can offer a finished feature in 1.0 instead of just syntax for a feature that is not implemented yet.
== Gtk General interface stuff (all) ==
Do we still have interface problems like toolbars not showing up or menus needing two clicks, or anything of the sort, on any platform ? Is #195 still reproducible ?
I'm not aware of anything specific right now that prevents usage.
In general there's still a fundamental "glitchiness" to gtk3 on Windows, but I'm afraid that's just how gtk3 works on Windows these days. We'll certainly get some questions about it, but I wouldn't know what to do to significantly improve this until the release.
== Packaging (RdH, Patrick, others) ==
Approximately how much work is still needed for macOS stable dmg production or windows packaging ?
What happens with the ubuntu bionic "dependency wait" ? ( https://launchpadlibrarian.net/437056488/buildlog.txt.gz )
Updating our existing Windows packaging for 1.0 should be easy (once I can motivate myself to do it).
I always planned to look into cpack and investigate if it can produce proper installers, so we might eventually get rid of our own packaging and reduce the overhead of maintaining it, but that might need some more work.
== Performance (all) ==
Do we still have these massive performance problems as soon as the Objects dialog (or XML ?) has been opened ?
I'm not aware anybody pushed a fix? Didn't retest, though.
Apart from that performance stil lseems worse than 0.92.x (at least on Windows), but also here I wouldn't really know what to do. It's usable, but it's not good.
One particular performance issue (I know there were multiple reports, but not sure we have an actual issue for it and GitLab throws 500 errors at me): Selecting a new font takes a very long time (a few seconds at least, some take 10-20 seconds). Not sure if this is exclusive to Windows? It does not happen in 0.92.x, though (with the same version of fontconfig/harfbuzz/pango/...), so it's either something we're doing, or something gtk3 is doing differently from gtk2. I think performance wasn't always bad in master, though. I think it started either with the variable fonts support or after the SPStyle refactor.
== Text (Tav, Jabier, others) ==
How usable is the text tool ? Do we implement the "hamburger" toolbar ? Do we still have issues with variable font sliders ? How understandable are our text tools, and are users able to "just write text" without surprises ? (with line heights for instance ?)
Tbh, I can't use the text toolbar myself anymore... Either I'm to stupid or we really *should* do something about usability.
It never behaves as expected and always sets properties in ways that were not intended.
One line-height related issue that also needs attention is https://gitlab.com/inkscape/inkscape/issues/244
== SVG2 (Tav, all) ==
I'm not sure we have discussed enough the transition between SVG1.1 and SVG2. If we start producing svg2 documents that are not svg1.1-valid, should we use version="2" in <svg> ? how will it be understood by other renderers ? When and how do we intend to switch the 1.2 flawed text into the newly flawed text, how will we handle old documents, and how do we ensure they will render exactly the same ? Do we provide a way to save in "valid SVG1.1 document without SVG2 stuff" ?
I'd say we should always produce SVG1.1 fallbacks, even if we use SVG2 features?
Maybe we can then add an option to disable the compatiblity shims and produce a "pure" SVG2 document.
I know some switches are available, but Tav can probably comment better on what is implemented and what isn't.
== Documentation (Maren) ==
I saw https://gitlab.com/inkscape/inkscape-docs/documentation/issues/2 , do we have more stuff to update ? By the way, how do we sync the inkscape/inkscape repo with the inkscape/inkscape-docs ?
As Maren mentioned already there's a lot to update content-wise.
I've handled most of the technical side and meant to make a call for help at some point, but didn't get the chance to do it yet.
Once content is updated I typically copy the CI-generated files manually and push the update to the inkscape repo. Basically this can wait till shortly before 1.0, but we can certainly schedule a preliminary update for the beta(s).