[Search for users] [Overall Top Noters] [List of all Conferences] [Download this site]

Conference azur::mcc

Title:	DECmcc user notes file. Does not replace IPMT.
Notice:	Use IPMT for problems. Newsletter location in note 6187
Moderator:	TAEC::BEROUD

Created:	Mon Aug 21 1989
Last Modified:	Wed Jun 04 1997
Last Successful Update:	Fri Jun 06 1997
Number of topics:	6497
Total number of notes:	27359

781.0. "INFO: Uniqueness of Multi-word Attribute names" by COOKIE::KITTELL (Richard - Architected Info Mgmt) Mon Mar 11 1991 15:28

The necessity of parsing command lines justifies a somewhat stronger SRM
statement of uniqueness of attribute names relative to child entity names.

To whit: multi-word attribute names must not rely on the second and
subsequent words of their name to disambiguate the attribute from a
child of the same entity.

For example, given

	GLOBAL ENTITY Branch Library

	  ATTRIBUTE Cart Name Cache Count
	  ATTRIBUTE Task Persistence Time
	  CHILD ENTITY Cartridge
	  CHILD ENTITY Task

the attributes can't be addressed via FCL, because "Task Persistence" could be
an attribute name or a reference to the child Task with a name of "Persistence".
If the parser always looked for an attribute name first, then it could handle
a reference to the attribute but a Task entity with a name of "Persistence" 
could never be referenced.

With the V1.1 MCS kit, references to the attributes bring a warning, with
an unexpected attribute name.

MCC> show branch aim Task Persistence Time
%MCC-W-ATTRUNKNOWN, unknown attribute TASK

I would have expected it to try and find an attribute Time on a child Task
called Persistence, and say unknown attribute TIME.

With the iconic interface there aren't any syntactic ambiguities, so this
isn't an issue. But you might consider having the next rev of the SRM
mention this little gotcha in the section that deals with the uniqueness
of names.

T.R	Title	User	Personal Name	Date	Lines
781.1		TOOK::CALLANDER		`Wed Mar 13 1991 13:22`	24
	Actually you are right on target...but the reason you didn't see what you expected was because the parser tried multiple things before giving you the error message. MCC> show branch aim Task Persistence Time We recognized the Verb no problem; then we were able to pick up branch aim; we then pick up task and find an entity match and attempt to match on the instance (I don't know what data type task has but it looks like persistence didn't match) if the entity didn't match it would then try to match "task" to an attribute of branch, and that is the error you saw. If "persistence" had been a legal instance value on task I would have expected an error on "time". Now I could be off some where because I made a number of assumptions here that might be invalid (like task even takes an instance); but I think you get the general idea. Please note though there is still a qar on the FCL for it's lack of informative error messages. jill
781.2	instance wouldn't match	COOKIE::KITTELL	Richard - Architected Info Mgmt	`Wed Mar 13 1991 17:56`	5
	Jill, your assumption was correct, the identifier on Task is numeric, so Persistence wouldn't match. I guess the presence of the Task entity causes the parser to focus on Task, instead of trying to match Task Persistence Time with the attributes.
781.3	Still a bit confused...	DFLAT::PLOUFFE	Jerry	`Wed Mar 13 1991 19:24`	28
	RE: .1 > if the entity didn't match it would then try to > match "task" to an attribute of branch, and that is the > error you saw. Jill, I still don't understand something. The parser seemed to do all the right things up to the moment after it tried to "match "task" to an attribute of branch". Why didn't it try to continue and see is "task persistence" was a valid attribute of branch? Assuming that this would fail, why didn't it continue to see if "task persistence time" was a valid attribute of branch? In other words, why did the parser fail to realize that "task", "persistence" and "time" were separate keywords of a valid attribute of the branch class? It seemingly was trying to parse for an attribute and the parser already parses multi-keyword attributes. I'm just curious... - Jerry P.S. I have a hunch that it really tried to parse for an attribute starting with the word "persistence". Is this possibly what happened?
781.4	more detail on parsing	TOOK::HAO		`Thu Mar 14 1991 12:30`	23
	Hi Richard, I don't know why your command is failing. From what I understand of the parser, it should be doing something close to what Jill described. When the parser gets to the keyword "Task", it tried to match on it as a child entity of "Branch". When that failed due to the instance datatype mismatch, the parser tries to match "Task" as the first keyword of an attribute. IF that succeeded, the parser checks if "Task" has another keyword. The parser keeps trying to parse the longest (set of keywords) attribute it can find, as indicated by the parse tables. If the longest attribute isn't valid for the entity, the parser will backtrack -- lop off the keywords to try and find if a shorter attribute is valid. However, judging from your error msg-- "unknown attr" -- it sounds as if the keyword "Task" was NOT found in the parse tables at all. This is different from finding "Task" in the parse tables but not knowing that it was valid for your entity. You may want to doublecheck that your parse tables are properly updated. Christine
781.5	INFO: check of parse tables	COOKIE::KITTELL	Richard - Architected Info Mgmt	`Thu Mar 14 1991 13:56`	13
	Hi Christine. > However, judging from your error msg-- "unknown attr" -- it sounds as > if the keyword "Task" was NOT found in the parse tables at all. This > is different from finding "Task" in the parse tables but not knowing > that it was valid for your entity. You may want to doublecheck that > your parse tables are properly updated. From the MCC_PTB_PARSER.DAT file: Command: SHOW BRANCHLIBRARY TASK PERSISTENCE TIME
781.6	last suggestion	TOOK::HAO		`Fri Mar 15 1991 12:10`	12
	Hi Richard, Did you try specifying "show branch library aim task persistence time" instead of leaving the 2nd keyword of the GE out? Do you get the same error or a different one? At this point, I'm out of suggestions. My knowledge of FCL and parsing grows fuzzier every day!! Anyways, Jill or Pete Ditmars can help you from here. Christine
781.7	single-word global entity name	COOKIE::KITTELL	Richard - Architected Info Mgmt	`Fri Mar 15 1991 13:34`	13
	RE: .6 > Did you try specifying "show branch library aim task persistence time" > instead of leaving the 2nd keyword of the GE out? Do you get the same > error or a different one? Sorry, I entered the global name incorrectly in the basenote. We had to change the name from Branch Library to BranchLibrary for now because the iconic PM doesn't find the icon file for multi-world global entities. So it is a one-keyword entity name, and "Branch" is just an abbreviation of it. Thanks for the suggestions, Christine.