failed with setting simple grid

[petebau]

Hi there,
pls excuse if I should be too stupid...


I'm on version 3.29.4.0. Community ed. Linux 64bit.
To reproduce the effect, no drawing was neccessary.

I've tried to simply set a grid for snapping and easier viewing.

grid 0.125m
meta grid 1m

Works as expected when zoomed in, to approx. 1.5 metres displayed horizontally.
Behaves weird when zoomed out, to 2 metres or more displayed.

The line in the last screenshot snapped to the correct grid on the right end (point set when zoomed in) and snapped to the weird grid on the left end (point set when zoomed out a bit).

Why does it switch the grid to ca. 155mm instead of the 125mm I asked for?


Thanks. kind regards
Pete

[CVH]

Hi, and welcome to the QCAD forum

This issue is well known and described here:
https://www.qcad.org/rsforum/viewtopic.php?f=33&t=10155
Not OS nor QCAD version related, it is part of the initial commit or at least 11 years old.

Your settings also count as the minimum grid setting.
Zooming in it will not drop below the values that you have set.
But when zooming out something weird happens.
Your setting is first squared to 0.15625 units and zooming out further it will be increased 10 based:
1.5625 - 15.625 - 156.25 - 1562.5 and so on.

It is a little shame that this issue is not addressed in over 2 and a half year.
Especially because the culprit is documented and a solution is simple.
It boils down to bad math.
Not really convinced that 'a user defined list of all acceptable grid spacing's' as Andrew proposes is a solution.

For now you need to stick to 10 based values as listed in the drop down box.
That won't allow for 1/8 * 10^n meter increments for the grid points, sorry.

Regards,
CVH

[petebau]

Oooh no ... I see.
Thanks for the explanation (and for your previous efforts towards a correction).

Is there a chance that someone will fix the code soon? I admit that I wouldn't dare to even look there.


As for the circumvention:
Unfortunately my hardware is not capable of 16000 by 10000 pixels.

Meanwhile I could, of course, draw anything 8 times bigger than real, so that the error does not kick in yet for useful zoom settings. Would be a good math training for my brain, at least. But I'm still too slow:).


Great creators of qcad, please hear my prayer!

[CVH]

As for the circumvention:
Unfortunately my hardware is not capable of 16000 by 10000 pixels.

QCAD is vector based so there are in fact no pixel based limits.
How that is presented on paper in any DPI depends on a paper scale later on.
To work with bitmap files is not an issue because 1pixel can be scaled freely to units and/or back.

draw anything 8 times bigger than real, so that the error does not kick in yet for useful zoom settings.

Grid = 1 & MetaGrid = 8 won't hold either:
Minimal would be 1 < 8 units.
Next steps are: 10 < 8 ; 10 < 0 ; 100 < 0 ... and so on while reminding that 1, 10, 100 are actually 0.125, 1.25, 12.5
Zero means that a MetaGrid is not displayed.
Basically not workable IMHO.

Regards,
CVH

[petebau]

I thought... more pixels, more zoom possible, before the automatic spoils everything. Ok, good consolation that expensive hardware wouldn't have helped.

But, without any math thoughts, just a feeling, I had tried the "8" and "1" and it worked ways better than 1000 / 125.


btw.,,
if someone else should need to draw something in 125mm grid or similar in the meantime:

Disabling the automatic grid scaling in Application Preferences/Graphics View (why not a Drawing Preference?) switches off the weird error, instantly. In this mode, the grid just disappears when too fine. And -important- the min. number of pixels for grid spacing can be set, hence no broken grid and a useable grid snap in this mode. Oof.

Thanks a lot