Status updates - Felipe Sanches - Inkscape

[Felipe Sanches]

Hello, everybody!

My name is Felipe Corrêa da Silva Sanches and I live in São Paulo, Brazil. I have been involved in vector graphics design software since my acceptance as a core developer in the Inkscape dev-team back in 2007. I participated in a couple Google Summer of Code projects including one in which I contibuted typography features in Inkscape. Because of that I met Dave Crossland. Four years ago, during a period of approximatelly 1 year we worked together on these typography features in Inkscape and now, (after a 3-year hiatus in which I opened and run my own 3d printer manufacturing startup here in Brazil) I'm back working with Dave again. :-D

In the past months I worked in developing improvements in Metapolator, including import and export routines for the OpenType and UFO file formats. From now on I'll be also working again on Inkscape improvements, including in-person meeting with Marcelo Magalhães to gather feedback on usability issues of PowerStroke and any other things that may be useful for his type design efforts in Inkscape.

My daily status updades have been done on the Metapolator G+ community here:
https://plus.google.com/communities/110027004108709154749
But now I'll also post updates to this mailing-list whenever the topic is not related to Metapolator.

I'm preparing to travel tomorrow to the Chaos Communication Camp in Berlin, Germany. So I'll be away until August 18th and you can expect my next status update to be sent to this mailing list on August 19th, then :-)

=====

After contacting the local Brazilian community to gather feedback on what type-design features they would want to see in Inkscape, I got these suggestions:

* The ability to export font files directly from Inkscape in formats such as ttf and otf.
* A crazy idea of a wizard for semi-automactically extracting glyph curves from a scanned image (by tracing the original image and allowing the user to map each portion of the vector result to a specific unicode codepoint)
* The ability to query libre font repositories from within Inkscape (something similar to the Open Clipart Library integration)
* another user also suggested a feature to output in font file formats such as woff/ttf/otf/svg
* a user mentions that he feels that Inkscape lacks the ability to embed fonts into a document

this last user also mentioned something fuzzy about relative paths that I really didn't understand.

Happy hacking,

Felipe "Juca" Sanches

PS:

I also blog (mostly about unrelated topics) here:
http://mamedev.emulab.it/fsanches/

And here: http://jucablues.blogspot.com.br/

[Felipe Sanches]

I just got back from Berlin. The Chaos Communication Camp 2015 was a great hacker community gathering and I'm proud I was there again this year! I'm still a bit jet-lagged, but I was able to resume work today on typography stuff for Inkscape/Fontforge.

Today I had a video-chat (using WebRTC) with Marcelo Magalhães to catch up with the work he's been doing on tamil script while I was away. Based on our 2+ hour long video-chat and some further investigation on the afternoon we've got a few questions sorted out.

Marcelo is working with Inkscape and Fontforge in order to design his latin & tamil glyphs. One of the issues he presented was the fact that fontforge has been crashing out-of-the-blue this week. He's been testing several different versions of fontforge and all of them seem to crash in spurious/random fashion. One work-around he found was to move to a different computer. By using a different computer, the same fontforge build seems to behave better. We suspect it may be an issue with the X11 layer, but we're not sure.

The set of glyphs that Marcelo is expected to implement during this project is described in this file:
https://github.com/adobe-fonts/source-tamil-sans/blob/master/GlyphOrderAndAliasDB

He's got some trouble to load this glyph-map file into fontforge in order to setup an initial work-place for him with this specific subset of unicode codepoints. A user called Tharique Azeez has helped Marcelo here in this mailing list by providing a postscript file which supposedely contains the map of all tamil unicode codepoints that he'll have to draw. But we're not sure how one could generate such a postscript file based on the original GlyphOrderAndAliasDB, or vice-versa. Or even validate if both files are describing equivalent unicode ranges. Are there even specific tools for that?

Marcelo asked me if I would like to see the guides from this website translated to portuguese: http://designwithfontforge.com/en-US/index.html Well... I think it would be great, but I don't think I'm the best person to be involved in such effort as I believe I would be more useful to the project by contributing code or perhaps a final review of the documentation in order to provide developer-oriented technical feedback.

Finally, I've been recently very annoyed by a UI bug in the Inkscape PowerStrokes feature: Whenever we finish dragging, the control points of the LivePathEffect have their positions updated on the canvas at once (not smoothly during the drag). But I feel this is pretty much irritating and I would consider it much better if the control nodes did move along while the user is draging so that the user can get immediate visual feedback. That's something that made me dive into the live path effects source code in Inkscape looking for a way to improve this behaviour while also getting a better understanding of the underlying code structure.

I'll be glad to get feedback from anybody who's got any insight on these issues. I know this is a vague email covering lots of topics at once. I'm just getting back to work and I'm not sure yet exactly which of these issues deserve priority on my next work days.

cheers,

Felipe

[Felipe Sanches]

Oh! I also should mention that when I suggested Marcelo should be using the very latest nightly build of Fontforge he informed me that the link in the website is broken:
http://fuuko.libferris.com/osx/packages/FontForge_latest-HEAD.app.dmg

I wonder if people simply stopped building and publishing these fontforge packages every night, or if it is instead due to some other broken code in the fontforge website.

[Felipe Sanches]

Today's work:

* Recovered my username/password credentials on launchpad.net

* Registered a new publish SSH key on my launchpad account (the one I use on my current computer)

* With that, I was able to push commits to the Inkscape project using git-bzr

* This was the simple commit I pushed for testing purposes: http://bazaar.launchpad.net/~inkscape.dev/inkscape/trunk/revision/14314

* Then I started investigating this crasher bug in PowerStrokes: https://bugs.launchpad.net/inkscape/+bug/1487424

The crasher bug provides an example file that is supposed to cause the issue. But I noticed that Inkscape hangs always, when a powerstroke is applied to a path. The other Live Path Effects are working nicelly.

The user ~suv identified that the issue has been introduced last week when 2geom was updated in this commit:
https://bazaar.launchpad.net/~inkscape.dev/inkscape/trunk/revision/14299

Tomorrow's work:

* Attempts to trace and fix the issue

* Communicate with the 2geom developers

[Dave Crossland]

On 20 August 2015 at 20:57, Felipe Sanches <ju...@members.fsf.org> wrote:

But we're not sure how one could generate such a postscript file based on the original GlyphOrderAndAliasDB, or vice-versa. 

Use the https://github.com/adobe-type-tools/afdko - docs at http://www.adobe.com/devnet/opentype/afdko.html

Or even validate if both files are describing equivalent unicode ranges. Are there even specific tools for that?

I'm not sure how  Tharique created the file. 

However, Frank Trampe put a lot of work into the UFO support in FontForge last year, such that it could be used as a UFO editor in order to compile fonts using the afdko. I think that using the afdko is better than using FontForge to do so. 

[Felipe Sanches]

This Sunday:
* studied 2geom math concepts and corresponding source code
* while focusing on fixing https://bugs.launchpad.net/inkscape/+bug/1487424

[Felipe Sanches]

Today I continued investigating the Powerstroke bug (https://bugs.launchpad.net/inkscape/+bug/1487424) There are 2 suggested fixes. One was suggested by me and the other one was presented by a user called Alvin Penner. Neighter of these two proposed fixes are convincing to me. I feel that I need to understand the implementation in more detail before settling on a fix for the bug.

So I spent some time reading the PowerStroke code and sketching geometry on paper in order to understand the linear algebra involved in the implementation. Power Stroke hangs Inkscape very frequently (and in unpredictable ways). It definitely needs a thorough code review to improve its quality.

I also had a video call with Marcelo Magalhães today in which we discussed in more detail the issues he's been facing while using Inkscape and Fontforge. One major issue that we identified is that Marcelo (and potentially other users as well) wants to setup his workplace in fontforge with the set of unicode codepoints that cover Latin-1 and the Tamil script, so that he can see all glyphs at once. The project scope specification refer to an encoding map describer by this file: https://github.com/adobe-fonts/source-tamil-sans/blob/master/GlyphOrderAndAliasDB We decided that fontforge should support parsing the GlyphOrderAndAliasDB file format when it is loaded via the Encoding->Load Encoding menu option. We may work on implementing support for that in fontforge.

[Felipe Sanches]

Today:

* Forked on github and build Fontforge from source

* cleaned up code style of the routine that handles loading of encoding files

* prepared the code to parse a new file format (detect by filename if the user is trying to load a GlyphOrderAndAliasDB file)

* boilerplate code for parsing GlyphOrderAndAliasDB

Tomorrow:

* Implement the actual parsing routine for GlyphOrderAndAliasDB file format

* chat with Marcelo to map other usability issues on Inkscape and Fontforge

[Dave Crossland]

On Aug 25, 2015 9:16 PM, "Felipe Sanches" <ju...@members.fsf.org> wrote:
>
> Today:
> * Forked on github and build Fontforge from source
> * cleaned up code style of the routine that handles loading of encoding files
> * prepared the code to parse a new file format (detect by filename if the user is trying to load a GlyphOrderAndAliasDB file)
> * boilerplate code for parsing GlyphOrderAndAliasDB
> Tomorrow:
> * Implement the actual parsing routine for GlyphOrderAndAliasDB file format

Well, I think this is not a good idea. I think it's better to make a js script you can drop a GlyphOrderAndAliasDB file into and get out a font forge Nam file. But if you think it will only take 1 more day, better to finish what you started.

> --
> You received this message because you are subscribed to the Google Groups "Google Fonts Discussions" group.
> To unsubscribe from this group and stop receiving emails from it, send an email to googlefonts-dis...@googlegroups.com.
> To post to this group, send email to googlefon...@googlegroups.com.
> Visit this group at http://groups.google.com/group/googlefonts-discuss.
> To view this discussion on the web visit https://groups.google.com/d/msgid/googlefonts-discuss/CAK6XL6BCq0-o6o7VNfcWsE6p6hUBsGpEf3Rj8uVsSjPhUXnhSQ%40mail.gmail.com.
>
> For more options, visit https://groups.google.com/d/optout.

[Dave Crossland]

On Aug 24, 2015 10:02 PM, "Felipe Sanches" <ju...@members.fsf.org> wrote:
>
> Today I continued investigating the Powerstroke bug (https://bugs.launchpad.net/inkscape/+bug/1487424) There are 2 suggested fixes. One was suggested by me and the other one was presented by a user called Alvin Penner. Neighter of these two proposed fixes are convincing to me. I feel that I need to understand the implementation in more detail before settling on a fix for the bug.
>
> So I spent some time reading the PowerStroke code and sketching geometry on paper in order to understand the linear algebra involved in the implementation. Power Stroke hangs Inkscape very frequently (and in unpredictable ways). It definitely needs a thorough code review to improve its quality.
>
> I also had a video call with Marcelo Magalhães today in which we discussed in more detail the issues he's been facing while using Inkscape and Fontforge. One major issue that we identified is that Marcelo (and potentially other users as well) wants to setup his workplace in fontforge with the set of unicode codepoints that cover Latin-1 and the Tamil script, so that he can see all glyphs at once.

If you load a file like Pria's Catarmaran then you can save a name list and load it in a fresh project. Or just select all and copy and clear the fore layer to back; then you have working OpenType to work with too

> --
> You received this message because you are subscribed to the Google Groups "Google Fonts Discussions" group.
> To unsubscribe from this group and stop receiving emails from it, send an email to googlefonts-dis...@googlegroups.com.
> To post to this group, send email to googlefon...@googlegroups.com.
> Visit this group at http://groups.google.com/group/googlefonts-discuss.

> To view this discussion on the web visit https://groups.google.com/d/msgid/googlefonts-discuss/CAK6XL6Ag4uVp_7DuwMJzVOhyntO%3DXbkEuwmD%2BpY15%2BvJKqd8OA%40mail.gmail.com.

[Pathum Egodawatta]

But Fontforge nam file only stores names of the Unicode glyphs. Non Unicode glyphs are not on the nam file.

To do what you suggest, it has to be a Encoding.ps file it will show up as a custom Encoding.

To view this discussion on the web visit https://groups.google.com/d/msgid/googlefonts-discuss/CAEozd0y52zrZ%2BSPRhKa13H-aKpLSN%2BOCzwZAnWfz1rxO5oxdFQ%40mail.gmail.com.

[Dave Crossland]

On Aug 25, 2015 11:40 PM, "Pathum Egodawatta" <pat...@mooniak.com> wrote:
>
> But Fontforge nam file only stores names of the Unicode glyphs. Non Unicode glyphs are not on the nam file.

Give them pua encodings for development.

> To do what you suggest, it has to be a Encoding.ps file it will show up as a custom Encoding.

Whatever works without writing fontforge c code is good.

> To view this discussion on the web visit https://groups.google.com/d/msgid/googlefonts-discuss/CAHEFZc5fewiKW7pn70x31f6_5Gt6hi4CsPq45pa0sYum%2Ba5qkQ%40mail.gmail.com.

[Pathum Egodawatta]

I use a custom Encoding file for this.

Here are the files I've been using
https://github.com/mooniak/fontforge-encodings

I think it's the easiest way to manage.

GOADB support for FF is something I really prefer, my workflow is a nightmare if I change the glyph set halfway. Since Encoding.ps is only storing names of non Unicode glyphs. So everyone you need both Encoding.ps and nam files to maintain a proper set up. Phrasing GOADB would do the job.

[Felipe Sanches]

Today:
* more cleanup and refactoring of fontforge C code related to parsing encoding files

* implemented parsing of GlyphOrderAndAliasDB

* tested it with the sample file provided by Marcelo and it works

* code is available at https://github.com/felipesanches/fontforge/tree/ParseEncodingFile
* A pull request has been opened here: https://github.com/fontforge/fontforge/pull/2467

Tomorrow:
* chat with Marcelo to map other usability issues on Inkscape and Fontforge [postponed from today]

* more coding of whatever we come up with on the video chat ***OR*** some more work on Metapolator export code (rewiring it to Jeroen's recently merged new UI code)

[Dave Crossland]

Sounds great :) 

[Pathum Egodawatta]

On Thu, Aug 27, 2015, 9:13 AM Felipe Sanches <ju...@members.fsf.org> wrote:

Today:
* more cleanup and refactoring of fontforge C code related to parsing encoding files

* implemented parsing of GlyphOrderAndAliasDB

* tested it with the sample file provided by Marcelo and it works

* code is available at https://github.com/felipesanches/fontforge/tree/ParseEncodingFile
* A pull request has been opened here: https://github.com/fontforge/fontforge/pull/2467

Awesome! We can ship it for the next release! I'll test.

Tomorrow:
* chat with Marcelo to map other usability issues on Inkscape and Fontforge [postponed from today]

* more coding of whatever we come up with on the video chat ***OR*** some more work on Metapolator export code (rewiring it to Jeroen's recently merged new UI code)

--

You received this message because you are subscribed to the Google Groups "Google Fonts Discussions" group.
To unsubscribe from this group and stop receiving emails from it, send an email to googlefonts-dis...@googlegroups.com.
To post to this group, send email to googlefon...@googlegroups.com.
Visit this group at http://groups.google.com/group/googlefonts-discuss.

To view this discussion on the web visit https://groups.google.com/d/msgid/googlefonts-discuss/CAK6XL6Dfe3_1vV-9W90_qqN1v_KX6TFk%2B_MuRhOQhwrnx5X0bA%40mail.gmail.com.

[Felipe Sanches]

Thanks, Pathum!

Be sure to let me know if you perceive any issue with the current implementation. I'll be glad to fix any eventual glitches.

happy hacking,

Felipe Sanches

To view this discussion on the web visit https://groups.google.com/d/msgid/googlefonts-discuss/CAHEFZc6sJXkoYcB3qkb4m-LXC9%3DeMON%2BtKP%3DPAH8Q4Sntkc3iA%40mail.gmail.com.

[Felipe Sanches]

Today:

* a bunch of code cleanup/refactoring in fontforge

* Video-chat with Marcelo. We talked a lot about the Coiny project that Marcelo is developing, including considerations about the project deadline and potential future projects, but also a bit about the drawing tools' technical challenges and about the upcoming ATypI conference here in São Paulo, Brazil, when we'll meet Dave in person again ;-)

Tomorrow:

* Metapolator export code

[Felipe Sanches]

Today I re-synced my local working copy of Metapolator with upstream master, adding the several dozens of commits developed by Jeroen along the past months. As far as I understand it, his code improvements finally bring all of the development of Metapolator strictly to the master branch. Previously the UI code was developed in the gh-pages branch while only the Metapolator engine was kept on master.

After updating my local copy, though, it was unclear how to bootstrap the new code. Previously I used to run a local webserserver (python -m SimpleHTTPServer) and access the Metapolator webapp at localhost:8000/purple-pill Now I'm not sure exactly how to do that or which folder the app actually resides in now. I asked it to Lasse Fister, but did not get an answer from him yet. So I'll simply continue doing some more work on Fontforge today.

I have some strong feelings about Fontforge, actually. The source code is a mess and sometimes I think that Dave's reasoning about the ideal programming languages for type-design software that resulted on the choice of Javascript (instead of C/C++) for a next-gen font tool such as Metapolator, may be a bit biased by how badly designed fontforge is in terms of its source code architecture and code-style.

I do think of the source code of a program as an integral part of the program. Being a software freedom activist I trully believe that software without its corresponding source code is fundamentally incomplete. And thus, I believe that just as we have UI usability issues, we also have "source code usability" issues, in the sense that poorly written code makes it less effective to be useful and extensible for its users.

For that reason, I insist that we must not let fontforge's source code be abandoned in its poor indentation & code style, but instead we should revitalize it, even if resorting to automated tools such as GNU indent, so that it can get less cryptical and thus more easily improved. Along the way, we'll surely find bugs and fix them, since there are very weird and non-obvious constructs that are hidden in the mess of the current code.

[Dave Crossland]

On 28 August 2015 at 19:55, Felipe Sanches <ju...@members.fsf.org> wrote:

I have some strong feelings about Fontforge, actually.

So do its many ex-users ;)

 

The source code is a mess and sometimes I think that Dave's reasoning about the ideal programming languages for type-design software that resulted on the choice of Javascript (instead of C/C++) for a next-gen font tool such as Metapolator, may be a bit biased by how badly designed fontforge is in terms of its source code architecture and code-style.

That's not why Lasse chose JavaScript. https://github.com/metapolator/metapolator/wiki/faq#why-is-metapolator-written-in-javascript-not-python explains.

 

I do think of the source code of a program as an integral part of the program. Being a software freedom activist I trully believe that software without its corresponding source code is fundamentally incomplete. And thus, I believe that just as we have UI usability issues, we also have "source code usability" issues, in the sense that poorly written code makes it less effective to be useful and extensible for its users.

For that reason, I insist that we must not let fontforge's source code be abandoned in its poor indentation & code style, but instead we should revitalize it, even if resorting to automated tools such as GNU indent, so that it can get less cryptical and thus more easily improved. Along the way, we'll surely find bugs and fix them, since there are very weird and non-obvious constructs that are hidden in the mess of the current code.

The lead developers in the FF community reject running indent because it tramples git blame metadata that is more useful for working on the code than indentation style. The reason it is cryptic is not the whitespace. 

I have spent the last 3 years and significant funds to improve FontForge, and it is a black hole. It is not worth development beyond crash fixes and tweaks that take a few hours at most to implement. Your encoding update is the largest kind of change I would support, and even then I think writing Python scripts would be a better investment. 

If you want to raise funds for FontForge and work on it, good luck! :) I would much prefer you to contribute to the new defconqt once it is published. I hope it will be soon.

--

Cheers
Dave

[Felipe Sanches]

Tell us more about defconqt :-)

--

You received this message because you are subscribed to the Google Groups "Google Fonts Discussions" group.
To unsubscribe from this group and stop receiving emails from it, send an email to googlefonts-dis...@googlegroups.com.
To post to this group, send email to googlefon...@googlegroups.com.
Visit this group at http://groups.google.com/group/googlefonts-discuss.

To view this discussion on the web visit https://groups.google.com/d/msgid/googlefonts-discuss/CAEozd0yhciK49udPjdpiBo_qO5ssjrTr9-MenomAvdwr1LBLvg%40mail.gmail.com.

[Dave Crossland]

On 28 August 2015 at 20:56, Felipe Sanches <felipe....@gmail.com> wrote:

Tell us more about defconqt :-)

Adrien Tétar was one of the most active FontForge developers since George Williams quit the project, but he ran out of patience with it after about a year, and started porting https://github.com/typesupply/defcon from Apple's Cocoa to PyQt5. You can see screenshots at https://twitter.com/adrientetar/media . He hasn't published the source code yet but I don't know what he is waiting for. 

[Adrien Tétar]

2015-08-29 3:00 GMT+02:00 Dave Crossland <da...@lab6.com>:

On 28 August 2015 at 20:56, Felipe Sanches <felipe....@gmail.com> wrote:

Tell us more about defconqt :-)

Adrien Tétar was one of the most active FontForge developers since George Williams quit the project, but he ran out of patience with it after about a year, and started porting https://github.com/typesupply/defcon from Apple's Cocoa to PyQt5.

Also, I did not “run out of patience”–the months I spent working on the project gave me the technical insight to decide what the best way forward was (that is what an engineer usually does btw, 1/ analysis and 2/ development).

Generally speaking I do not work with patience (or lack thereof), I just try to find the best paths to attain my goals.

 

You can see screenshots at https://twitter.com/adrientetar/media . He hasn't published the source code yet but I don't know what he is waiting for. 

Adrien

[Dave Crossland]

On 29 August 2015 at 04:39, Adrien Tétar <adri....@gmail.com> wrote:

2015-08-29 3:00 GMT+02:00 Dave Crossland <da...@lab6.com>:

On 28 August 2015 at 20:56, Felipe Sanches <felipe....@gmail.com> wrote:

Tell us more about defconqt :-)

Adrien Tétar was one of the most active FontForge developers since George Williams quit the project, but he ran out of patience with it after about a year, and started porting https://github.com/typesupply/defcon from Apple's Cocoa to PyQt5.

Also, I did not “run out of patience”–the months I spent working on the project gave me the technical insight to decide what the best way forward was (that is what an engineer usually does btw, 1/ analysis and 2/ development).

Generally speaking I do not work with patience (or lack thereof), I just try to find the best paths to attain my goals.

I apologies for putting words in your mouth. https://meta.wikimedia.org/wiki/Cunningham%27s_Law strikes ;)  

[Felipe Sanches]

Today:
* Some fontforge troubleshooting with Marcelo and a video chat with him as well
* There's a feature request that he's got for Inkscape that I was not
able to understand, so we'll revisit that face-to-face
* I spent some more time on the LPE sources

Tomorrow:
* There's some Metapolator work that´s been waiting in the queue for a
while. I gotta give it some attention tomorrow afternoon.
* And I'll meet Marcelo at his office.

> --
> You received this message because you are subscribed to the Google Groups
> "Google Fonts Discussions" group.
> To unsubscribe from this group and stop receiving emails from it, send an
> email to googlefonts-dis...@googlegroups.com.
> To post to this group, send email to googlefon...@googlegroups.com.
> Visit this group at http://groups.google.com/group/googlefonts-discuss.
> To view this discussion on the web visit

> https://groups.google.com/d/msgid/googlefonts-discuss/CAEozd0wcS%3D%2BsBMAftcB4%3Dhhec5TzON9gB%3D5O4Lfjm1Of8LKXew%40mail.gmail.com.

[Felipe Sanches]

Today I met Marcelo at his office. He showed to me the work he's been
doing on FontForge, designing the Coiny font. The attached photos
include some sample text printed on paper using the Latin glyphs and
lots of Tamil glyphs in the FontForge workplace on his laptop screen
(only a few of the Tamil ones are actually seen in the pictures,
though). But one could check everything by checking out the git
repository and loading the sfd file:

https://github.com/marcelommp/Coiny/

Having the opportunity to talk to him face to face was good
(complementing out almost daily video-chats). Now I finally understand
what Marcelo is looking for as an improvement to the Inkscape
NoteTool. He wants a path edit mode in which the handles of the
neighbour path segments are adjusted proportionately when you edit a
node's position. This way, he can more easily tweak glyph outlines.
One example of how that could be useful would be when creating other
weigths to an existing font. Naturaly, Marcelo is working on a single
weigth for Coiny right now. But even then, being able to easily adjust
the weigth of a glyph is important when trying to reach a good weigth
balance between all glyphs in a font.

Right now I won't be working on this tool improvement because I plan
to focus the most of my time on Metapolator in the hope of having a
really nice demo for the upcoming ATypI conference:
http://www.atypi.org/atypi-2015-sao-paulo-brazil-1

Also, today I'm happy that there's a new thread in the Inkscape
mailing list about the measure tool that I created with Dave 4 years
ago. There's people willing to improve it with snapping features and
some other things:
http://sourceforge.net/p/inkscape/mailman/message/34424580/

[Dave Crossland]

On 1 September 2015 at 22:03, Felipe Sanches <ju...@members.fsf.org> wrote:

He wants a path edit mode in which the handles of the
neighbour path segments are adjusted proportionately when you edit a
node's position.

I made a vide of this in FF a while ago, https://www.youtube.com/watch?v=q2mNOTo9lqQ