Hi,
Actually I wrote the code similar to what you guys are saying. Then Dave B pointed to me what he was trying to do. Initially I said wow that will be nice. I might need that instead of putting on the comments of ad_page_contract how to properly use the included .tcl.
I went into my merry way hoping that someday ad_page_contract will be like that. Anyway after a while due to extensive use of includes and the forms and lists that I was making, I keep on forgeting what parameters to pass. So I have to read the comments I put in the child file. Then I remember what DaveB told me, I asked DaveB about that patch. I then applied it to my instance. I then found out that it did not work when the request came from a rp_internal_redirect. I fixed it a bit, then I ended up with the implementation above, after discussion with DaveB.
What this new behaviour means is that it enforces ad_page_contract on includes too. Actually the coding style of what Rob said is what I have done initially. I then told DaveB I don't get it. It works for me. Then he explained further, then I realize that DaveB idea of using includes and ad_page_contract is proper.
Should the includes pages blindly process the form vars. Or should the parent page do it? What also happens if the parent page process the form var? I am not sure if its ok to use another construct that functions the same as ad_page_contract.
Just look at this:
parent.adp
<include src="child" foo="bar">
child.tcl
ad_page_contract {} {foo:notnull}
In the current form of ad_page_contract. It will put an error of two values of foo. So the work around is either remove foo="bar" or remove foo:notnull. Which I think both workarounds seems to remove how the include is to be used. Or the include can't be used as stand alone, no ad_page_contract.
What this patch does is that. If a child.tcl is included. You have to follow what is stated in ad_page_contract which is foo:notnull and pass it in the include tag. Also with the patch its now possible to have the parent code to compute for foo. Which normally may be just a form var of child. This way pages that normally process form vars can be included on parent pages and the parent pages can alter and/or produce the vars in ad_page_contract.
ad_page_contract can now enforce passing through form vars or what vars much be present on include tags. I hope things are more clear. Hope to get more opinion about this. Thanks.
Come to think of it I maybe should call set_page_form_data_gathered 1 when processing includes.
