Re: DOCBOOK-APPS: clarify profiling logic
ed nixon wrote: is that, in order to be *included* in the profiled output, both attribute value conditions must be true; not one or the other. Yes, or attribute must be empty or not used. For example, if I have two sections as follows: section audiencelevel=management contentlevel=summary;detail ... /section section audiencelevel=staff contentlevel=summary;detail ... /section then I think I should be able to differentiate between these two sections with the following two parameter settings: audiencelevel=management contentlevel=summary;detail versus audiencelevel=staff contentlevel=summary;detail Using audiencelevel here is sufficient. The behaviour I'm getting from the profile.xsl stylesheet in the profiling directory for the latest version of the sheets seems to be giving me a logical OR rather than a logical AND. Weird. Am I misunderstanding the behaviors? Confused in my logical design? Only problem I see is that you are using nonexistent parameters. There are no parameters like audiencelevel and contentlevel, there are not even those attributes. You should use some DocBook attribute (e.g. userlevel, condition, ...) and then use parameters profile.userlevel, profile.condition and so on. What is the intended behavior if one of the profiling attribute values is null? When attribute is null then corresponding element goes to output. Jirka -- - Jirka Kosek e-mail: [EMAIL PROTECTED] http://www.kosek.cz
Re: DOCBOOK-APPS: clarify profiling logic
Jirka Kosek wrote: ed nixon wrote: snip/ The behaviour I'm getting from the profile.xsl stylesheet in the profiling directory for the latest version of the sheets seems to be giving me a logical OR rather than a logical AND. Weird. Am I misunderstanding the behaviors? Confused in my logical design? Only problem I see is that you are using nonexistent parameters. There are no parameters like audiencelevel and contentlevel, there are not even those attributes. You should use some DocBook attribute (e.g. userlevel, condition, ...) and then use parameters profile.userlevel, profile.condition and so on. Ok, so profile behavior of attributes added via DTD customization is undefined unless I were to customize profiling-mode(?) to take them into account? I assumed that simply naming the attribute with xsl:param name=profile.attributeaudiencelevel/xsl:param and then initializing it with xsl:param name=profile.valuemanager/xsl:param would do the trick. However, how the value got associatedwith the attribute under this scenario seems a little magical, now that I think about it. Although things seemed to work fine until I started to use combinations. How multiple value attributes came together would be even bigger magic, I guess. I'll try this with different, non-custom attributes and see where I end up. Thanks for your help. ...edN
RE: DOCBOOK-APPS: clarify profiling logic
You could also use a two-phase approach (it's what we are using). First profile with e.g. audiencelevel and then profile the result with contentlevel. Peter -Original Message- From: ed nixon [mailto:[EMAIL PROTECTED]] Sent: Mittwoch, 19. Februar 2003 19:21 To: Jirka Kosek Cc: docbookapp Subject: Re: DOCBOOK-APPS: clarify profiling logic Jirka Kosek wrote: ed nixon wrote: snip/ If you really think that you need more then one custom profiling attribute, then submit RFE. It would be easy to add parameters like profile.attribute1 and profile.value1. Thanks, Jirka. I can live with what there is now. Besides, I get a little twitchy when I see things like something1 and then something2 and then... I'm trying to think how a more general solution would look to the user. For example, xsl:param name=profile.attribute attname=audiencelevelmanager/xsl:param or xsl:param name=profile.attribute attname=osunix/xsl:param etc. But that's without thinking at all about what has to go on under the hood or whether it's a good idea in terms of consistency of style with the rest of the suite. Thanks again. ...edN
Re: DOCBOOK-APPS: clarify profiling logic
Peter Kullmann wrote: You could also use a two-phase approach (it's what we are using). First profile with e.g. audiencelevel and then profile the result with contentlevel. That's option, albeit very ineffective. -- - Jirka Kosek e-mail: [EMAIL PROTECTED] http://www.kosek.cz