[Date Prev] | [Thread Prev] | [Thread Next] | [Date Next] -- [Date Index] | [Thread Index] | [List Home]
Subject: RE: [sdd-comment] VersionRangeType
I see that I need to further explain the assumptions underlying my objections.
There are conceptually two distinct types of maintenance that are intended to be modeled in the SDD by Version and Fix Name. One type modifies a comparable version. Versions are known to be greater or less than each other. If you have a greater version, you know that it incorporates the contents of a lesser version. The other type of maintenance applies a named fix. Fix names only tell you that a particular fix is applied. You can't make any assumptions about what other fixes are applied. These distinct types exist in the realm we are modeling and the concepts need to be reflected in the SDD. To further clarify these concepts, imagine a simple world where comparable versions are numeric - like 1.2.3.2 and fix names are things like "Fix for Problem XYZ" and "Fix for Problem ABC". We intuitively understand that 1.2.3.2 is larger than 1.2.3.1 and incorporates the content of 1.2.3.1. It is also easy to understand from these names that just determining that the "Fix for Problem XYZ" is installed doesn't tell you anything about whether or not the "Fix for Problem ABC" is installed.
Our difficulty right now is that the actual realms that we need to model don't make things this obvious from the names of their maintenance packages. We have to dig a little to determine if a particular thing is really a version - and therefore comparable with greater and less than logic; or if it is really a fix name that is only comparable with equality. Windows makes this hard, but not impossible. The proposed solution would do away with the ability to model the named fix concept in the SDD by overloading its semantics. We need to dig deeper until we understand how Windows should map to these concepts - not throw out a necessary capability of the SDD.
I do have ideas about how Windows really does map to these concepts, but we need to agree on the concepts before that discussion will be useful.
Julia McCarthy
Install Strategy and Development
julia@us.ibm.com
877-261-0391
Jason Losh <Jason.Losh@sas.com>
02/19/2008 11:58 AM |
|
This proposal will not accomplish what you want. Fix name can only be compared for equality. Let's use your example of a min version of Windows 2000 with SP1 and max Version of Windows Vista with SP1. If the actual os is Windows 2000 with SP4, all you know is that it is not Windows 2000 with SP1. You do NOT know if it is within the required range. SP1 and SP42 are fix names and the semantics of fix name do not involve comparison for greater or less than. All you can know is that SP1 does not equal SP4.
Julia McCarthy
Install Strategy and Development
julia@us.ibm.com
877-261-0391
Jason Losh <Jason.Losh@sas.com>
02/18/2008 04:33 PM |
|
[Date Prev] | [Thread Prev] | [Thread Next] | [Date Next] -- [Date Index] | [Thread Index] | [List Home]