Where communities thrive


  • Join over 1.5M+ people
  • Join over 100K+ communities
  • Free without limits
  • Create your own community
People
Activity
    Worlddue
    @Worlddue
    Are there any jobs for these kind of computations?
    DeepSOIC
    @DeepSOIC
    i never looked. I was once invited to work on a cad package "Kompas 3d" (a pretty powerful russian parametric and do-it-all cad). But I wasn't ready to go.
    DeepSOIC
    @DeepSOIC
    I invite you all to discuss equal-quaternion constraint, https://forum.freecadweb.org/viewtopic.php?f=8&t=42899
    Worlddue
    @Worlddue
    Sorry been busy lately
    Going to look at it today
    Worlddue
    @Worlddue
    Hey quick question. When you are referring to external geometry in the solver what does that mean exactly?
    DeepSOIC
    @DeepSOIC
    Hi! I'm away now
    External geometry looks like any other geometry to sketcher, just the parameters are fixed
    Worlddue
    @Worlddue
    parameters fixed means that they are not dependent? Can only have 1 value?
    after being constrained?
    DeepSOIC
    @DeepSOIC
    not dependent??
    parameters that are fixed are simply hidden from the solver. The solver doesn't vary them because it doesn't know it can.
    Worlddue
    @Worlddue
    hm? Then what are fixed parameters? Can you give me any practical example?
    abdullahtahiriyo
    @abdullahtahiriyo
    No, it is a warning, but there are way too many warnings. If we keep flooded with warnings, then when there is an actual warning that is relevant to unveil a problem, we won't be able to detect it.
    DeepSOIC
    @DeepSOIC
    it is now flooding me with signed-unsigned stuff. I don't really know, what to do about it
    abdullahtahiriyo
    @abdullahtahiriyo
    Now you mean with my changes or without them?
    @DeepSOIC I have force push my branch with all my changes about warnings:
    on top of yours
    DeepSOIC
    @DeepSOIC
    I see you cleaned up a lot of warnings. Some are irrelevant (e.g., in ParaParabola, which has the placeholder implementation copied from ParaLine, i.e. completely irrelevant code =) )
    DeepSOIC
    @DeepSOIC
    I am quite reluctant to changing most ints to unsigned...
    DeepSOIC
    @DeepSOIC
    about all these unused kwd arguments. They are all legit warnings in some sense: if kwd is ignored, it means I can give the constructor random keyword arguments, and they will all be accepted and completely ignored. That is not right, so there should be a check on them.
    so i'm wondering, should all this warning cleanup be postponed for later? Because cleaning them up now is silencing some legit warnings...
    this code is very very unfinished after all
    abdullahtahiriyo
    @abdullahtahiriyo
    There are two ways to silence an unsigned int to signed int. Unless you want negative indices (there is one such case), the preferred way is using size_t instead of int. If you need negative indices, then casting the unsigned to signed is the only way.
    Being the code unfinished should not be a reason to allow all types of warnings. You will fail to see problems (potentially errors in the sense of mistakes that compile).
    With respect to kwd, then the check code should be added. It is acceptable that the check does not handle any that maybe will be handled in the future, but otherwise we get flooded with warnings.
    abdullahtahiriyo
    @abdullahtahiriyo
    With respect to the unused parameters associated with throwing an exception because it is not implemented. There is no point in seeing these warnings if the underlying issue is that is not yet implemented. I think it is better to silence them, because when the implementation comes, they will be made available again if needed.
    DeepSOIC
    @DeepSOIC
    If I had the power, I'd choose to make std::vector use ssize_t instead of size_t, and forget about all this unsigned nonsense. Unfortunately, I don't have the power =(

    With respect to the unused parameters associated with throwing an exception because it is not implemented.

    of course, we should just silence them.

    DeepSOIC
    @DeepSOIC
    i'm not too worried about being flooded with warnings at compile time, because qt creator provides me this:
    image.png
    DeepSOIC
    @DeepSOIC
    Unless you want negative indices (there is one such case),
    there is one big such case, in ParameterSubSet + ValueSet object combo. ParameterSubSet has a lookup table, which convers indexes. And for the parameters not in the subset, it has -1-es. This is relied upon in multiple places in ValueSet.
    ///Lookup table. Input = index of parameter in store. Output = index of parameter in the set.
    std::vector<int> _lut;
    DeepSOIC
    @DeepSOIC
    Also, yes, (lack of) warnings are valuable. But so is code readability. If it takes a lot of typecasts just to silence warnings, I would prefer to disable the problematic warnings instead, for the whole file.
    DeepSOIC
    @DeepSOIC
    I've fixed some unused kwd by actually using them.
    abdullahtahiriyo
    @abdullahtahiriyo
    DeepSOIC
    @DeepSOIC
    pushed some code
    DeepSOIC
    @DeepSOIC
    oops, I messes up some commits, will force-push soon
    DeepSOIC
    @DeepSOIC
    force-pushed.
    DeepSOIC
    @DeepSOIC
    @abdullahtahiriyo i'm trying to rebase, getting some merge conflicts that are not straightforward to solve. Can you offer some help?
    image.png
    DeepSOIC
    @DeepSOIC
    This is my first attempt at rebasing. Haven't even tested if it builds or not.
    DeepSOIC
    @DeepSOIC
    I hope I have sorted it out by myself. DeepSOIC/FreeCAD-ellipse@9d51478