FIX: check bounds for fill->min_spacing causing fill_density<100 to abort

[ElectroQuanta]

Fill->min_spacing was being initialized to 0, due to the infill_extrusion_widthbeing set to 0 in the config file.

• As such, I defined a lower bound value to be positive (>0), in spite thinking that this should be at least the size of nozzle_diameter.
• Now, I don't get any exceptions as discussed here.

Tests

For convenience, I had the result of some tests:

• Infill_extrusion_width = 0.5:
  
• Infill_extrusion_width = 0.1:
  
• Infill_extrusion_width = 2.1:
  
• Infill_extrusion_width = 0.01: Was taking too long.

This lasts two values highlight the importance of convenient bounding, but for now, I will leave as it is, as it suits my needs.

[lordofhyphens]

Revert this change.

[lordofhyphens]

I will take a closer look over this tomorrow and (probably) commit it. I apologize for not getting to it from now.

[ElectroQuanta]

No need to apologize :)

Like I said, I commited to the wrong branch, as time-stamping is used for my debugging and cross-referencing. Nonetheless, I guess it would not harm, if it could be added upstream (maybe as transversal feature).

With that said, I wouldn't know why that raised the red flag, because the travis output below:

src/libslic3r/SLAPrint.cpp: In member function ‘std::string Slic3r::SLAPrint::get_time() const’:
src/libslic3r/SLAPrint.cpp:356:10: error: ‘put_time’ is not a member of ‘std’
    ts << std::put_time(std::localtime(&t), "%c %Z");
          ^
At global scope:

mentions wrong scope, but it complies to my compiler settings. From the same travis output, I see that gcc 4.9 is being used, causing it to fail.

I could rewrite to a more portable/compatible version, but like I said it was just for my debugging :)

Sorry for any inconvenience :) Always learning :)

[lordofhyphens]

I've made some changes to your PR to instead set a relatively sane value for the config width.

You'll still get credit for the merge (and spotting the issue in the c++ CLI toolchain). I made a similar fix on the perl side a few months ago (that I just remembered when I saw your PR).

The root cause here is that the config code assumes that the infill width gets calculated as per flow (which doesn't exist here).

I think we will want, going forward, to refactor Flow so that the auto logic behaves appropriately for SLA/DLP/. I use a LCD mask on my own printer.

[ElectroQuanta]

Yeah, I also thought that proper bounding was the way to go :)
I just didn't know the proper way to bound it :)

The way I see it (and I could be wrong), but if the fill_spacingis less than nozzle_diameteryou'll get overlapping on the filling, and thus, the bounding suggested would not suffice :)

But, like you said, if the problem is upstream (Flow) and you already identified the need to refactor it, for me it's good, because like I said, it suits my needs for now :)

And thanks, for further investigating the issue :)

[lordofhyphens]

In that code path, nozzle_width doesn't exist ;)