Number: 294

Date: 26-Mar-84 13':34':42

Submitter: Sannella.PA


Subject: SINGLEFILEINDEX fails to list some recordtypes

Assigned To: Jonl


Status: Closed

In/By: Harmony

Problem Type: Bug

Impact: Annoying


Frequency: Intermittent

Priority: Perhaps

System: Other Software




Lisp Version: 

Source Files: 

Microcode Version: 

Memory Size: 

File Server: 

Server Software Version: 

Disposition: '
[Date': 12 Apr 84 23':43 PST'
Subject': AR 294'
Fixed on <LispCore>LIbrary>, as per my comment of Apr 3.]'
["Sannella.PA" "22-Aug-84 17':43':25" Attn': Status':(Fixed->Closed) In/By':]

Description: '
Date': 25 Mar 84 15':44 PST'
Subject': Lisp': Singlefileindex'
Lisp-System-Date': 15-Mar-84 00':13':18'
Machine-Type': Dorado'
Fails to list a records entry for recordtypes not on the usual clisprecordtypes.  '
-- Tayloe.'
Date': 29 MAR 84 11':42 PST'
From': JONL.PA'
Subject': Re': AR 294': SINGLEFILEINDEX fails to list records'
To':   Sannella, Jonl'
cc':   Stansbury'
In response to the message sent  26 Mar 84 13':36':36 PST (Monday) from Sannella.PA'
See the documentation on teh extensions to SINGLEFILEINDEX -- the'
variable DEFAULTINDEXEDTYPESLST holds the "driver" for what things will'
be indexed.  I think the documentation is clear enough for a user to add'
to the entry on DEFAULTINDEXEDTYPESLST under "RECORD".  If not, then the'
documentation should be upgraded.'
Date': 29 Mar 84 14':41 PST'
Subject': [JONL.PA': Re': AR 294': SINGLEFILEINDEX fails to list records]'
To': Jonl,'
Yes, changing DEFAULTINDEXEDTYPESLST will do the trick, but that is unnecessarily inconvenient. '
I would argue that DEFAULTINDEXEDTYPESLST ought to allow for indirection, and indirect the kinds of records indexed to the current value of CLISPRECORDTYPES.  '
-- Tayloe.'
Date':  2 Apr 84 21':00 PST'
AR 294 -- I decline.  Further comments': An automatic interface with '
        "indirection" as suggested in this AR is not feasible (in general)'
        since SINGLEFILEINDEX is taking a rather limited, local-context'
        look at the expressions on the file while it is trying to parse it'
        looking for definitions.  The documentation mentions this "heuristic"'
        approach at finding the definitions.  I really speeds things up to '
        do it this way (orders and orders of magnitude!)'
Date':  3 APR 84 17':37 PST'
From': JONL.PA'
Subject': AR 294 re-vivified!'
To':   LispSupport'
cc':   Stansbury, Tong'
Yesterday I "declined" this AR, but after talking it'
over with Tayloe, it indeed does appear to be the case that this particular'
instance will work with "indirection" -- the condition must be that the'
"name" for the definition must be exactly the name for the function which'
MAKEFILE uses to output the definition on the file.'
Thus, e.g., ASSOCRECORD has this property, since that is the name for the'
kind of record and its definition will appear on a fil as'
(ASSOCRECORD <foobar> ...)'
whereas BITMAPS won''t work because BITMAPS is the filepkg name for the'
definition, but it is output on a file by MAKEFILE as'
(RPAQ <foobar> ...)'
In light of all this, I''m attempting to add this "indirection" feature'
to SINGLEFILEINDEX, for records to indirect through the value of'
CLISPRECORDTYPES, to be in the next version on [Phylum]<LispCore>Library>.'
There is a protocol that governs what goes on CLISPRECORDTYPES, and it'
gurantees the abovementioned property.'


Test Case: 

Edit-By: Sannella.PA

Edit-Date: 22-Aug-84 17':43':26