How to set the order of a PCell's Parameters?

 Hi,

Yes, the CDF parameter names must match the pcell parameter names.  Instance level CDF is more specific than the cell CDF that you would create for the the parameters, so these values should be passed down to the pcDefinePCell statement.  If you are using IC615 then the SKILL IDE can handle PCell SKILL code so you might debug the issue that way?  Otherwise, you can create labels in the PCell code or use global variables for better visibility of what is going on.

Regards,

Lawrence.

Thank you both for the help. I'm running into a different problem now-
My parameters in the pcell definition are as follows:

(
(NX int 1)
(NY int 2)
(B1 boolean "FALSE")
(B2 boolean "FALSE")
)

I'm using the code from here to generate the cdf. It works great, except the ordering is incorrect. If I take the reverse() out, the booleans are in the correct order, but I end up with:

NY, B1, B2, NX

Instead of the correct order. Looking at the cellview: cv~>parameters~>value~>name shows the same, incorrect order. I can fix this by manually coding the CDF, but I'd like to avoid that. Any ideas why it's throwing off the order?

 Thanks.

Edit: I hacked around this by manually adding the cdf for car(reverse(cv~>parameters~>value)), then running the foreach() loop on reverse(cdr(reverse(cv~>parameters~>value)). This will work assuming the first one is always shifted last for some reason. I'd still appreciate some input though!

 I wrote the Solution that you are referring to. I'm not sure that the 'boolean' parameters are coded correctly?  Shouldn't it be:

(
(NX "int" 1)
(NY "int" 2)
(B1 "boolean" nil)
(B2 "boolean" nil)
)

?  I have not tried this, but if you can try changing the code and see if it helps.  Boolean parameters are a bit awkward if I recall correctly, since the way that they are treated in CDF is not the same as the pcDefinePCell statement.

I hope that this helps!

regards,

Lawrence.

 Gave that a shot, but it didn't work. It seems that its an issue with having multiple types in there because if I only have booleans it works fine.

OK, I took a look at the solution, and it is assuming that the parameters hierProp is ordered. I don't think there is anywhere which says that this is the case - that the order matches the order the parameters were defined in the pcDefinePCell. You might have made that empirical observation in IC5141, but it was never guaranteed - and with the change in the database to OA, it doesn't appear to be the case any more (I didn't try, I'm just basing this on what you've observed). Regardless of whether it is true or not, it is not a valid assumption to have made.

That said, it would be easy to modify the code in Lawrence's solution to define the order without "manually" coding it. Sure, you have to define the order, but apart from that it would be straightforward:

    foreach(param foreach(mapcar paramName '("NX" "NY" "B1" "B2") dbFindProp(cv~>parameters paramName))
      ;; create the cdf in the same basic way, but depending on the
      ;; parameter do some things differently
      cdfCreateParam(cdf
      ...

So you are just explicitly listing the order you want when you do the CDF parameter creation.

Hope that helps,

Regards,

Andrew.

Note - if you can't see all the code above because it's truncated in your browser, selected it and copy/paste into an editor.

I see what you mean. If I do a cv~>parameters~>value~>name in the IC615 CIW, the parameter order comes up in the strange way where the first one defined is last (i.e. NY B1 B2 NX). In IC5141, however, it comes up reversed, i.e. B2 B1 NY NX, thus that solution would work.

Is there any way to set the hierProp order? In the meantime, I'll try the code you posted; thanks.

Edit: The code works great. Thanks again for the help.