<html>
<head>
<style><!--
.hmmessage P
{
margin:0px;
padding:0px
}
body.hmmessage
{
font-size: 12pt;
font-family:Calibri
}
--></style></head>
<body class='hmmessage'><div dir='ltr'><div>Let's try to schedule a meeting for this week or next.</div><div>My calendar this week is not as busy as it usually is. Feel free to suggest and I will block the time. </div><div>It always get busier as it gets in to the week. </div><div>See my answers below. </div><div><br></div><div>> Hi guys,<br><br>> I have thought some more about the OSrL question of the last few <br>> exchanges. I do not have a solution, but there are other questions, <br>> about both OSrL and OSoL.<br><br><br>> 1. OSoL> We have <other> elements in <variables>, <constraints> and <br>> <objectives>, and there is a "type" attribute. What does that <br>> attribute refer to? I think it is ambiguous. Since the <var> array is <br>> optional, it seems that there could be an option something like<br>> <br>> <other name="relaxIntegrality" type="real" value="0.2"/><br>> <br>> In this case, the "type" refers to the value of the option. But in <br>> other cases, such as<br>> <br>> <other name="stageInfo" type="integer" numberOfVar="2"><br>> <var idx="0" value="1"/><br>> <var idx="1" value="2"/><br>> </other><br>> <br>> what does the "type" attribute of the <other> refer to? I assume it <br>> refers to the value inside the <var> element. Is that your reading <br>> also? </div><div>Yes. If other is an array, as in the above example, it is then an integer array. </div><div><br></div><div>>Should there then be an attribute "varType" and "enumType" to <br>> allow for the possibility that an <other> option has both a global <br>> value and a number of <var> children, with possibly different types? </div><div>No. The value should be either scalar or vector, but not both. I would assume~</div><div> <br>> Also, why does each <var> element have an optional "name", and should <br>> there be a restriction that either a <var> array or an <enumeration> <br>> can be given, but not both? Or should the restriction also cover the <br>> "value" attribute, that is, there can be at most (exactly?) one of <br>> "value", <var>, <enumeration>? In the latter case, we could get by <br>> with the single "type" element, but we had better document this!!!<br>I am not sure about the question exactly. </div><div>I also peeked at the emails ahead in the queue. We probably can talk in the meeting some time this week or next.<br><br>> 2. The same question for the OSrL schema.<br>> <br>> 3. In addition, when we set up "type" for OSrL, I think we should have <br>> an enumeration of types. What should go into that enumeration?<br>> <br>> integer<br>> real (or double?)<br>> char<br>> string<br>> boolean (? needed? "true"/"false" are strings)<br>> ?<br>R/B/I/S</div><div>for real/binary/integer/string</div><div><br>> Cheers<br>> <br>> gus<br>> -------------------------------------------------------<br>> <br>> Horand I. Gassmann, Professor<br>> <br>> School of Business Administration, Dalhousie University<br>> 6100 University Avenue, PO Box 15000<br>> Halifax, Nova Scotia, Canada, B3H 4R2<br>> ph. (902) 494-1844<br>> fax (902) 494-1107<br>> <br>> http://myweb.dal.ca/gassmann/<br>> <br>> _______________________________________________<br>> Os-project-managers mailing list<br>> Os-project-managers@list.coin-or.org<br>> http://list.coin-or.org/mailman/listinfo/os-project-managers<br></div> </div></body>
</html>