Another old approach that might work better in the
who-knows-how-long-the-user-will-take-or-even-if-they're-still-logged-in
world of the web is
1) when selecting include a magic field (explained below)
2) let user make edits with the UIl; when ready to save
3) within a db transaction check to see that the magic field has not
changed; if unchanged, do the update, but if changed, re-select data, notify
user, display their data and the just-retrieved data so they can decide to
overwrite or back off
SQL Server (Sybase & Microsoft) had/have a specific datatype for this
"magic" field called TIMESTAMP (a name that was a recurrent source of
confusion) but it could be implemented in any modern RDBMS; you just need a
field that, through triggers or something, is _guaranteed_ to increment any
time the record is updated.
> -----Original Message-----
> From: Gonzo Rock [mailto:[EMAIL PROTECTED]]
> Sent: Friday, August 31, 2001 6:39 PM
> To: CF-Talk
> Subject: Mulit-User Concurrency
>
>
> Sure...
>
> I see, Thanks Pete.
> A Cached Array for persistent data which is index pointers so
> to speak... yep makes sense.
>
> Ok... NOW, Thinking deeper...
>
> How are people handling Multi-User concurrency issues? How to
> not overwrite someone elses changes to a record while
> pondering how to make your own changes to the record....
>
> IE:
> User A reads Record #1.
> User B reads record #1.
> User B changes fields[3,4,5] in Record #1 and rewrites the
> entire record.
> User A changes fields[7,8,9] in Record #1 and overwrites
> [3,4,5] with old data just changed by User B.
>
> In the past I have used the Classic
> Lock/ReReadRecord/ModifyMyChangedFieldsOnly/WriteRecord/UnLock
> scenario to avoid this via built in tools...
>
> What does CF have built in or in the way of custom tags to
> help with this?
>
>
>
>
>
>
> At 04:01 PM 8/31/01 -0400, you wrote:
> >Gonzo,
> >
> > There are cases where building a cached array is more
> efficient. Mainly
> >when you want indexed Access to records (basically your not
> going to output
> >all your data at once).
> >
> >So consider the following situation lets say you had a table called
> >documents that you accessed the following way...
> >/getDocument.cfm?DocumentID=12 and it would pull the
> document from the db
> >with that ID out and display it. If you had a cached query
> your going to
> >have to loop through it and find the correct document (or do
> a query of a
> >query). It's much faster to access an array element here
> than perform this
> >comparison.
> >
> >Here's how you might build such an array, for this general
> document example:
> >
> >put this in application.cfm ...
> >
> >
> ><cflock scope="application" type="exclusive" timeout="20">
> ><cfif NOT IsDefined("application.document")>
> > <cfquery datasource="dsn" name="doc">
> > SELECT DocumentID, Name, Description, Content
> > FROM documents
> > </cfquery>
> >
> > <cfset application.document = ArrayNew(1)>
> > <cfoutput query="doc">
> > <cfscript>
> > application.document[DocumentID] = StructNew();
> > application.document[DocumentID].Name =
> doc.Name;
> >
> application.document[DocumentID].Description = doc.Description;
> >
> application.document[DocumentID].Content = doc.Content;
> > </cfscript>
> > </cfoutput>
> ></cfif>
> >
> ></cflock>
> >
> >
> >Now to access an element, something like this...
> >
> ><cflock scope="application" type="readonly" timeout="20">
> >
> <cfoutput>#application.document[url.DocmentID].content#</cfoutput>
> ></cflock>
> >
> >Note that if someone types in a DocumentID into the url that
> is not in the
> >array an error will be thrown, you need to catch that error.
> Also do not
> >use this technique to store large amounts of data, because
> your server could
> >run out of RAM.
> >
> >
> >++++++++++++++++++++++++++++++++++++++++++
> >Pete Freitag ([EMAIL PROTECTED])
> >CFDEV.COM
> >ColdFusion Developer Resources
> >http://www.cfdev.com/
> >
> >
> >-----Original Message-----
> >From: Gonzo Rock [mailto:[EMAIL PROTECTED]]
> >Sent: Friday, August 31, 2001 3:36 PM
> >To: CF-Talk
> >Subject: RE: Efficient Query Manipulation?
> >
> >
> >Bryan...
> >
> >Love that you responded... I was beginning to feel invisible
> on this list
> >;-)
> >
> >Ummmmmm... Ok You say the DBServer should absorb the load,
> not the CFServer.
> >
> >So Refinding 100 indexed database records is
> faster/easier/MoreResource
> >Efficient than CFServer manipulating 100 records stored in
> an Array. Very
> >Interesting point.
> >
> >RE the interface:
> >
> >Thanks for the ideas... not sure what I'll be doing...
> Currently I have
> >three submit buttons for each Row now... insert above this
> row, delete this
> >row, edit this row. Just want to work out the best way to do
> this before
> >going on because it will be central and used heavily in the
> rest of this
> >application.
> >
> >again, thanks for making me not feel so invisible ;)
> >Gonzo
> >
> >
> >
> >
> >At 12:17 PM 8/31/01 -0700, you wrote:
> >>It is not very efficient to convert query results to an array. The
> >overhead
> >>involved is completely avoidable. More often than not the
> database machine
> >>will be more powerful (and reliable) than the CF server.
> When you are
> >faced
> >>with a decision between loading up the DB machine or the CF
> machine I would
> >>choose the DB machine every time.
> >>
> >>You may also want to modify the interface...
> >>
> >>Place a checkbox next to each row containing that row's ID and have
> >multiple
> >>submit buttons at the bottom such as "Delete All Checked
> Rows", and "Move
> >>Checked Rows Up", and "Move Checked Rows Down". This way
> you can modify
> >>many rows with one query and you don't have to reload the
> page more than
> >>once. You would need to put a little thought into the
> process in order to
> >>maintain an un-broken sequence (i.e. how to avoid a gap
> between 5 and 7
> >when
> >>deleting 6) and so on...
> >>
> >>
> >>Bryan Love ACP
> >>Internet Application Developer
> >>Telecommunication Systems Inc.
> >>[EMAIL PROTECTED]
> >>
> >>
> >>
> >>-----Original Message-----
> >>From: Gonzo Rock [mailto:[EMAIL PROTECTED]]
> >>Sent: Friday, August 31, 2001 10:23 AM
> >>To: CF-Talk
> >>Subject: Efficient Query Manipulation?
> >>
> >>
> >>I am rather new at CF and am trying to decide how to go
> about manipulating
> >>sets of records in a database.
> >>
> >>1)Use cached queries
> >>-- OR --
> >>2)Build an array.
> >>
> >>My dilemma goes like this...
> >>
> >>- A user Queries the database and gets a set of say 50 to
> 100 records in a
> >>list. On of the fields retrieved is the SEQUENCE of the
> record in the list.
> >>- Next the user both inserts new items into and deletes
> current items from
> >>this list until they are satisfied with the contents of the list.
> >>
> >>
> >>
> >>I can update the database on each insert or deletion, modifying many
> >records
> >>to update the SEQUENCE value of the records. (An obvious
> Downside is lots
> >of
> >>database activity)
> >>
> >>-- OR --
> >>
> >>My most current thinking is that stuffing the Query into an
> Array would be
> >a
> >>good idea... Then I could manipulate the Array, inserting
> and deleting rows
> >>until a time where the user would commit the Array. (An
> obvious Downside is
> >>loss of all work done in the session if the
> App/OrSomethingElse crashes.)
> >>
> >>Is this an acceptable practice?
> >>
> >>Are there techniques?
> >>
> >>What's the view from the Brainiacs on this list?
> >>
> >
>
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
Your ad could be here. Monies from ads go to support these lists and provide more
resources for the community. http://www.fusionauthority.com/ads.cfm
FAQ: http://www.thenetprofits.co.uk/coldfusion/faq
Archives: http://www.mail-archive.com/cf-talk@houseoffusion.com/
Unsubscribe: http://www.houseoffusion.com/index.cfm?sidebar=lists
