Expand Stroke Needs Work

[Mithferion]

5 minutes ago, LarrySunshine said:

Yes but in prod where you cut stuff, foil etc there are just too many details so you can overlook easily even if you triple check. If you are holding your breath every time you launch prod worrying your software may have failed you it's just not the way to work. If you have to check on everything everytime then you may as well write your own code to operate plotters, printers etc. Keep in mind we are talking about critical bugs dev's know about for years. It's no excuse.

I know there is no excuse, and I’ve said that it has been badly managed by Serif.

Also, there are key points where checking should be done, not in every single step.

I remembered another weird stuff. When installing Oracle EPM System, if it’s a Virtualized environment, tell the Administrator to save a Snapshot before you click “Run” in the Installer. Why? Because it’s going to fail ay some point, and you will need to revert back to the Snapshot, and then, you will repeat the installation process some more times with the exact same steps and exact same configurations, and it will eventually work.

Frustrating, I know.

Best regards!

[Jowday]

1 hour ago, A_B_C said:

Insinuations … first if all, I had the impression that Jowday meant to insinuate that the developers at Serif are not sincere in their expressed commitment to improve Expand Stroke, which I found indeed a bit offensive towards the team behind Affinity Designer.

Wrong. I compared to the struggle Boris is going through because even though you know he is trying, the result is impossible to predict. Many variables at play.

So the fact alone that he is trying is not worth a penny. The result counts. And whatever users write and say, they are not a credible or interesting source.

The actual implementation RELEASED counts and it is overdue. 

[Jowday]

We are talking about pants on first, then shoes priorities here. Not optimizing the CERN accelerator complex.

[Oval]

591 days ago the shoes of the CERN accelerator complex were presented as follows:

Quote

its non-trivial to produce a mathematical offset from a cubic bezier that is correct. It is however possible to produce an offset that is ‘good enough’ that you’d never know it wasn’t right. We already have this code. I’ve said before that I intend to rewrite the expand strokes function to produce the actual offset curves in this fashion rather than re-fitting the expanded geometry as we currently do. I will implement this as soon as I’m able. To be clear, this is very important and will be done as soon as I can - but it is not trivial to implement and needs to be done carefully (ie not in a rush!) or the results will be just a different kind of wrong!

Not only repeating complaints, it was always made hope.

2015:

Quote

will soon be back on with looking at boolean ops, expand stroke

2016:

Quote

the Expand Stroke will be worked on very soon

Quote

At present there is no Contour Tool - this will be added in time. For the moment , you can produce the same effect (albeit with more effort!) by adding a stroke to your object and then choosing 'Layer'->'Expand Stroke'. Please accept my apologies for this inconvenience - this will be addressed in time

2017:

Quote

When I write the new version ofexpand stroke  it should essentially provide a very nice contour function at the same time

Quote

improve this to provide better results on thin strokes for the next beta...

2018:

Quote

The dev team is already aware of it but only rewriting/improving the code will fix it which may still take some time. 

 

[Patrick Connor]

1 hour ago, Jowday said:

Jebus, so sensitive....

Has it occurred to any of you that some people on this forum (indeed in this thread) may be more sensitive than you? 

I think that you have summarized the problem and Serif's responses adequately, and I do not think any more attacking or defensive posts will achieve anything.

I have always defended the rights of users like Oval, Jowday and others to hold our feet to the fire in public, but there is a thin line between questioning our responses and intentions and calling us liars and implying we don't care. I think this thread is best locked now before that line is encroached on further.

[MattP]

On 9/9/2019 at 9:54 AM, LarrySunshine said:

The size of element shouldn't matter, so you should obviously know as a dev. A proper vector editor has to be precise. It's actually even worse if the result is barely noticeable so if big prod is launched and the result doesn't fit, looks funny etc. I mean you know all this right?

LOL! Thanks for explaining things to me, very educating, I'll try to do better in the future. So if everything else is so accurate, do you want to explain to me what I'm seeing here?

(Blue background, white curve, semi-transparent pink result of expansion)

and, just for those commenters who are about to say "But what about CorelDraw? You didn't include it and it would've been perfect!"... (white background, black original curve, pink expanded result)

You'll notice that NOBODY has the right answer because - as I stated A LOT before - there IS NO RIGHT ANSWER because one curve does not perfectly become one other curve at an offset - MATHS DOESN'T WORK THAT WAY! It becomes potentially many curves, but in different ways. You need to use an algorithm to generate the offset and the results of the algorithm will vary.

So... sorry to break the awful news, but.... we're actually not that bad. Shocker. Double-shocker is the fact that I've actually shown you that we really do actually have a new stroke expansion function and the 'fanboys' aren't actually fanboys at all - they're just not ridiculously pessimistic keyboard warriors.

/micdrop