Groups | Search | Server Info | Keyboard shortcuts | Login | Register [http] [https] [nntp] [nntps]


Groups > comp.lang.clipper.visual-objects > #1626 > unrolled thread

Subclassing DbServer for auditing

Started byNick Friend <nicktekhne@googlemail.com>
First post2011-07-07 01:16 -0700
Last post2011-07-20 13:38 -0700
Articles 18 — 9 participants

Back to article view | Back to comp.lang.clipper.visual-objects


Contents

  Subclassing DbServer for auditing Nick Friend <nicktekhne@googlemail.com> - 2011-07-07 01:16 -0700
    Re: Subclassing DbServer for auditing "richard.townsendrose" <richard.townsendrose@gmail.com> - 2011-07-07 01:29 -0700
      Re: Subclassing DbServer for auditing Nick Friend <nicktekhne@googlemail.com> - 2011-07-07 01:56 -0700
    Re: Subclassing DbServer for auditing "Simon Goodman" <simon.spam@spam.abersoftware.co.uk> - 2011-07-07 13:25 +0100
    Re: Subclassing DbServer for auditing dlzc <dlzc1@cox.net> - 2011-07-07 09:01 -0700
      Re: Subclassing DbServer for auditing Nick Friend <nicktekhne@googlemail.com> - 2011-07-07 09:47 -0700
        Re: Subclassing DbServer for auditing dlzc <dlzc1@cox.net> - 2011-07-07 12:39 -0700
    Re: Subclassing DbServer for auditing "Geoff Schaller" <geoffx@softwarexobjectives.com.au> - 2011-07-07 22:15 +0000
      Re: Subclassing DbServer for auditing Nick Friend <nicktekhne@googlemail.com> - 2011-07-08 01:31 -0700
    Re: Subclassing DbServer for auditing Amilcar A. Camargo <amilcarcamargo@gmail.com> - 2011-07-07 17:33 -0600
      Re: Subclassing DbServer for auditing Nick Friend <nicktekhne@googlemail.com> - 2011-07-08 01:38 -0700
        Re: Subclassing DbServer for auditing dlzc <dlzc1@cox.net> - 2011-07-08 10:28 -0700
          Re: Subclassing DbServer for auditing "richard.townsendrose" <richard.townsendrose@gmail.com> - 2011-07-16 03:50 -0700
            Re: Subclassing DbServer for auditing Nick Friend <nicktekhne@googlemail.com> - 2011-07-18 00:24 -0700
              Re: Subclassing DbServer for auditing Benito Fernández <bfernandezm@prodigy.net.mx> - 2011-07-18 14:13 -0500
              Re: Subclassing DbServer for auditing Karl Faller <k.faller_withoutthat_@onlinehome.de> - 2011-07-18 21:17 +0200
              Re: Subclassing DbServer for auditing dengler <dewo001@hotmail.com> - 2011-07-20 21:29 +0200
                Re: Subclassing DbServer for auditing Nick Friend <nicktekhne@googlemail.com> - 2011-07-20 13:38 -0700

#1626 — Subclassing DbServer for auditing

FromNick Friend <nicktekhne@googlemail.com>
Date2011-07-07 01:16 -0700
SubjectSubclassing DbServer for auditing
Message-ID<e23586a3-110a-4a76-afcc-1cac477c5961@y13g2000yqy.googlegroups.com>
Hi all,

I need to retrofit a detailed auditing system to our application
(DBFCDX), and am looking for input...

We need to keep a register of changes at the record level, so for
instance if a user enters an editing screen and changes several fields
on one record, we should create a single register in the audit trail -
that register could either hold all the old field values and all the
new field values, or just the before and after values of fields that
have actually changed. The important thing is that it's a single audit
register for multiple field changes, not one audit record per field
changed.

I'm thinking the following at the moment (all in sub-classed
DbServer)..

In Append or Recall - set audit mode to "Create".
In Delete - backup all existing field values and set audit mode to
"Delete"
In FieldPut - if not already in audit "Create" or "Update" mode, then
must be modifying, so backup all existing fields and set audit mode to
"Update". If already in "Create" or "Update" mode, then don't need to
do anything.

In Unlock, Commit, Close, Skip, Goto, GoTop, GoBottom, Seek, Locate,
Continue - if in Create/Update/Delete mode, then complete audit of
record, backup all new field values and create audit register.

Some of this processing could probably be done in Notify, but looking
at the available Notify events there are some actions that aren't
really covered (eg. Recall) so it seems safer to handle it in the
individual methods.

Critical features are 1) registering changes at the record level (ie
grouping a single set of field changes to a record together), and 2)
obviously every database change must be caught and recorded.

Any thoughts/experiences?

TIA

Nick

[toc] | [next] | [standalone]


#1628

From"richard.townsendrose" <richard.townsendrose@gmail.com>
Date2011-07-07 01:29 -0700
Message-ID<cad2720e-2533-41a1-92ae-d7d5ad88157a@c41g2000yqm.googlegroups.com>
In reply to#1626
Nick

the only things you need to do are to store the changes when the save
button is hit

do you have packing and database maintenance ?

i would suggest:
save -> tableid, recordno, new/amend/delete, keyfield info, data
string old, data string new

packing would have to be done "manually"
when a table is packed then you have to redo the record nos. ouch ...

another thought: if you are running ads replication, can you not read
the updater table and store that ?

richard

[toc] | [prev] | [next] | [standalone]


#1629

FromNick Friend <nicktekhne@googlemail.com>
Date2011-07-07 01:56 -0700
Message-ID<270fb61a-461e-467c-a8e2-e05c0228a4f6@l18g2000yql.googlegroups.com>
In reply to#1628
Richard,

Two points...

First, I want to avoid having to go into existing code... there are
literally hundreds, probably thousands of places where data is changed
in the app, so I don't want to do that. Also if I miss one point where
data is saved in the code, the auditing is invalid. If I go to basics
(DbServer) I can catch this reliably throughout the whole app in an
automatic way as all data saving is eventually channeled via the
DbServer class. So I can do it by modifying only the DbServer subclass
and let the effects propagate through the system.

Second, I need to offer this to clients who don't have ADS, which is
most of my clients at the moment. ADS does this for replication...
effectively what I need to do is emulate that behaviour, but with
DBFCDX.

Nick

On Jul 7, 9:29 am, "richard.townsendrose"
<richard.townsendr...@gmail.com> wrote:
> Nick
>
> the only things you need to do are to store the changes when the save
> button is hit
>
> do you have packing and database maintenance ?
>
> i would suggest:
> save -> tableid, recordno, new/amend/delete, keyfield info, data
> string old, data string new
>
> packing would have to be done "manually"
> when a table is packed then you have to redo the record nos. ouch ...
>
> another thought: if you are running ads replication, can you not read
> the updater table and store that ?
>
> richard

[toc] | [prev] | [next] | [standalone]


#1632

From"Simon Goodman" <simon.spam@spam.abersoftware.co.uk>
Date2011-07-07 13:25 +0100
Message-ID<QAhRp.13010$T85.10813@newsfe19.ams2>
In reply to#1626
Hi Nick,

I had to do the same thing for my app a little while back.

I can send over the code I use - it's fairly generic as I tend to dump out 
the record before changes and then after changes to the file.

I did alter the code to call the audit functions though - just copy and 
paste at the relevant places but really tedious to do. Needed plenty of 
coffee and a deep breath before I started!

I'll dig out the functions - and send them via email - call me if you want 
to talk it through.

Simon


http://www.abersoftware.co.uk
"Nick Friend"  wrote in message 
news:e23586a3-110a-4a76-afcc-1cac477c5961@y13g2000yqy.googlegroups.com...

Hi all,

I need to retrofit a detailed auditing system to our application
(DBFCDX), and am looking for input...

We need to keep a register of changes at the record level, so for
instance if a user enters an editing screen and changes several fields
on one record, we should create a single register in the audit trail -
that register could either hold all the old field values and all the
new field values, or just the before and after values of fields that
have actually changed. The important thing is that it's a single audit
register for multiple field changes, not one audit record per field
changed.

I'm thinking the following at the moment (all in sub-classed
DbServer)..

In Append or Recall - set audit mode to "Create".
In Delete - backup all existing field values and set audit mode to
"Delete"
In FieldPut - if not already in audit "Create" or "Update" mode, then
must be modifying, so backup all existing fields and set audit mode to
"Update". If already in "Create" or "Update" mode, then don't need to
do anything.

In Unlock, Commit, Close, Skip, Goto, GoTop, GoBottom, Seek, Locate,
Continue - if in Create/Update/Delete mode, then complete audit of
record, backup all new field values and create audit register.

Some of this processing could probably be done in Notify, but looking
at the available Notify events there are some actions that aren't
really covered (eg. Recall) so it seems safer to handle it in the
individual methods.

Critical features are 1) registering changes at the record level (ie
grouping a single set of field changes to a record together), and 2)
obviously every database change must be caught and recorded.

Any thoughts/experiences?

TIA

Nick 

[toc] | [prev] | [next] | [standalone]


#1633

Fromdlzc <dlzc1@cox.net>
Date2011-07-07 09:01 -0700
Message-ID<27a26ed8-2e1b-4773-a987-5ede9121d6df@x10g2000vbl.googlegroups.com>
In reply to#1626
Dear Nick Friend:

On Jul 7, 1:16 am, Nick Friend <nicktek...@googlemail.com> wrote:
...
> Any thoughts/experiences?

How will you handle bulk updates, such as fields that might be
mathematically / rules derived from other fields?  Say price updates,
or something.

How about the production of temporary indexes, or changes of index
keys (the actual index order, not one key value)?

(x)Harbour has the ability to return the entire record
( recordinfo() ), including memo fields exploded out.  You probably
have something similar in VO.  A copy before and after edit will tell
you if something changed, and allow you to store the entire original
record if required.  Parsing such a beast might be more than I could
do...

I'd plan on adding date/time of edit, and logged in user name,
obviously.  Maybe computer name or IP address, as appropriate.

Will your plan cause the auditing database to generate a change record
for every change written to the auditing database?  So you might need
to select which databases get audited.

Had I known about this requirement ahead of time, I would have
performed all auditing on locally-maintained databases with single
records, being the record being edited.  That way, you can be a bit
more methodical about what is written where, when, and in what order.

Probably too obvious to be of any help.  No response required.

David A. Smith

[toc] | [prev] | [next] | [standalone]


#1634

FromNick Friend <nicktekhne@googlemail.com>
Date2011-07-07 09:47 -0700
Message-ID<42662bf9-2d3d-410e-941f-4705b54b96f9@j23g2000yqc.googlegroups.com>
In reply to#1633
David,

All comments are welcome ;-)

Bulk updates aren't really an issue as we don't use any commands like
AppendDB. As for cycling through a series of records, say updating a
field on a formula, the final write to the DBF will still be done by a
FIELDPUT at some point, and before a new record can be used a skip,
goto, seek, locate or suchlike has to be issued. So in the case of
recording updates to a series of records (perhaps with a filelock in
place), it would be a question of initiating an audit process at the
first fieldput (and record the curent state of the record), completing
the audit when a record movement takes places (so at that point we can
record the after state of the record and write everything to the audit
database), then record movement takes place, another FIELDPUT issued
to start writing to the next record, intitiate another audit process,
and so on.

This is why I want to tackle it at the DbServer level, because finally
all write processes will always have to be channelled through a small
number of methods there which can be intercepted.

The audit database itself would have to exempt from the process,
otherwise you end up with a recursive loop... so normal database files
would use derived audit-enabled DbServer class, audit file itself uses
plain DbServer.

This is how I'm seeing it.

Nick


On Jul 7, 5:01 pm, dlzc <dl...@cox.net> wrote:
> Dear Nick Friend:
>
> On Jul 7, 1:16 am, Nick Friend <nicktek...@googlemail.com> wrote:
> ...
>
> > Any thoughts/experiences?
>
> How will you handle bulk updates, such as fields that might be
> mathematically / rules derived from other fields?  Say price updates,
> or something.
>
> How about the production of temporary indexes, or changes of index
> keys (the actual index order, not one key value)?
>
> (x)Harbour has the ability to return the entire record
> ( recordinfo() ), including memo fields exploded out.  You probably
> have something similar in VO.  A copy before and after edit will tell
> you if something changed, and allow you to store the entire original
> record if required.  Parsing such a beast might be more than I could
> do...
>
> I'd plan on adding date/time of edit, and logged in user name,
> obviously.  Maybe computer name or IP address, as appropriate.
>
> Will your plan cause the auditing database to generate a change record
> for every change written to the auditing database?  So you might need
> to select which databases get audited.
>
> Had I known about this requirement ahead of time, I would have
> performed all auditing on locally-maintained databases with single
> records, being the record being edited.  That way, you can be a bit
> more methodical about what is written where, when, and in what order.
>
> Probably too obvious to be of any help.  No response required.
>
> David A. Smith

[toc] | [prev] | [next] | [standalone]


#1635

Fromdlzc <dlzc1@cox.net>
Date2011-07-07 12:39 -0700
Message-ID<95bd3473-6455-4555-8307-cc93db75db3f@r21g2000pri.googlegroups.com>
In reply to#1634
Dear Nick Friend:

On Jul 7, 9:47 am, Nick Friend <nicktek...@googlemail.com> wrote:
> David,
>
> All comments are welcome ;-)
>
> Bulk updates aren't really an issue as we don't use
> any commands like AppendDB. As for cycling through
> a series of records, say updating a field on a
> formula, the final write to the DBF will still be
> done by a FIELDPUT at some point, and before a new
> record can be used a skip, goto, seek, locate or
> suchlike has to be issued. So in the case of
> recording updates to a series of records (perhaps
> with a filelock in place), it would be a question
> of initiating an audit process at the first fieldput
> (and record the curent state of the record),
> completing the audit when a record movement takes
> places (so at that point we can record the after
> state of the record and write everything to the
> audit database), then record movement takes place,
> another FIELDPUT issued to start writing to the
> next record, intitiate another audit process,
> and so on.
>
> This is why I want to tackle it at the DbServer
> level, because finally all write processes will
> always have to be channelled through a small
> number of methods there which can be intercepted.
>
> The audit database itself would have to exempt
> from the process, otherwise you end up with a
> recursive loop... so normal database files
> would use derived audit-enabled DbServer class,
> audit file itself uses plain DbServer.

... or teach the superclass which tables are to be audited.  It surely
knows which alias the fieldput is applied to.

> This is how I'm seeing it.

I was just concerned with the number of records generated from a
"simple global change".  What could be found and potentially reversed
with a single audit record, will take one for every changed record.

Your way is formally correct, and has no holes.  Just seems like it
gets fat pretty quickly.

Good luck.

David A. Smith

[toc] | [prev] | [next] | [standalone]


#1636

From"Geoff Schaller" <geoffx@softwarexobjectives.com.au>
Date2011-07-07 22:15 +0000
Message-ID<4e16301d$0$2136$c30e37c6@exi-reader.telstra.net>
In reply to#1626
Nick,

You are pushing shit up hill. The only solution to your specifications 
is to subclass the DBServer class and trap every commit. You could 
certainly break into the Optimisticflush method (because it knows which 
fields where changed in the row it is updating) but be aware this audit 
file will scale to many times the size your original DBF. Very quickly. 
And it will be a significant performance impact.

But I have to say, this is where you should be using SQL.
This alone is worth the conversion.

Using triggers you can completely audit every event and transaction in 
your table system. With very little code impact and very low performance 
overhead.

Sorry, but SQL is just the way to go.

Geoff



"Nick Friend" <nicktekhne@googlemail.com> wrote in message 
news:e23586a3-110a-4a76-afcc-1cac477c5961@y13g2000yqy.googlegroups.com:

> Hi all,
>
> I need to retrofit a detailed auditing system to our application
> (DBFCDX), and am looking for input...
>
> We need to keep a register of changes at the record level, so for
> instance if a user enters an editing screen and changes several fields
> on one record, we should create a single register in the audit trail -
> that register could either hold all the old field values and all the
> new field values, or just the before and after values of fields that
> have actually changed. The important thing is that it's a single audit
> register for multiple field changes, not one audit record per field
> changed.
>
> I'm thinking the following at the moment (all in sub-classed
> DbServer)..
>
> In Append or Recall - set audit mode to "Create".
> In Delete - backup all existing field values and set audit mode to
> "Delete"
> In FieldPut - if not already in audit "Create" or "Update" mode, then
> must be modifying, so backup all existing fields and set audit mode to
> "Update". If already in "Create" or "Update" mode, then don't need to
> do anything.
>
> In Unlock, Commit, Close, Skip, Goto, GoTop, GoBottom, Seek, Locate,
> Continue - if in Create/Update/Delete mode, then complete audit of
> record, backup all new field values and create audit register.
>
> Some of this processing could probably be done in Notify, but looking
> at the available Notify events there are some actions that aren't
> really covered (eg. Recall) so it seems safer to handle it in the
> individual methods.
>
> Critical features are 1) registering changes at the record level (ie
> grouping a single set of field changes to a record together), and 2)
> obviously every database change must be caught and recorded.
>
> Any thoughts/experiences?
>
> TIA
>
> Nick

[toc] | [prev] | [next] | [standalone]


#1647

FromNick Friend <nicktekhne@googlemail.com>
Date2011-07-08 01:31 -0700
Message-ID<b81b869a-84d8-4da2-827c-320c3caae5fa@t5g2000yqj.googlegroups.com>
In reply to#1636
Geoff,

Obviously this would be easy in SQL, but this is not the question as
it needs to be done over DBFCDX. For us the aspect of the size of the
audit trail is not an issue, as the audit logs won't be left to
accumulate in the working directories, but will be shunted off
periodically by a background Windows service.

Nick

On Jul 7, 11:15 pm, "Geoff Schaller"
<geo...@softwarexobjectives.com.au> wrote:
> Nick,
>
> You are pushing shit up hill. The only solution to your specifications
> is to subclass the DBServer class and trap every commit. You could
> certainly break into the Optimisticflush method (because it knows which
> fields where changed in the row it is updating) but be aware this audit
> file will scale to many times the size your original DBF. Very quickly.
> And it will be a significant performance impact.
>
> But I have to say, this is where you should be using SQL.
> This alone is worth the conversion.
>
> Using triggers you can completely audit every event and transaction in
> your table system. With very little code impact and very low performance
> overhead.
>
> Sorry, but SQL is just the way to go.
>
> Geoff

[toc] | [prev] | [next] | [standalone]


#1642

FromAmilcar A. Camargo <amilcarcamargo@gmail.com>
Date2011-07-07 17:33 -0600
Message-ID<hefc1753qeho2909sq7h7nek1k5ld19rs3@4ax.com>
In reply to#1626
Hi Nick,

On Thu, 7 Jul 2011 01:16:06 -0700 (PDT), Nick Friend
<nicktekhne@googlemail.com> wrote:

>Hi all,
>
>I need to retrofit a detailed auditing system to our application
>(DBFCDX), and am looking for input...
>
>We need to keep a register of changes at the record level, so for
>instance if a user enters an editing screen and changes several fields
>on one record, we should create a single register in the audit trail -
>that register could either hold all the old field values and all the
>new field values, or just the before and after values of fields that
>have actually changed. The important thing is that it's a single audit
>register for multiple field changes, not one audit record per field
>changed.

I've done something similar in the past and this were the steps:

1. Subclass your dbserver from a master class. In this master class you will
add variables and methods. One variable holds the description or coding for the
modifications found. Other variable can hold a list (array) of field to track.
One more to disable tracking when it's needed. And a last one to tell if
something has changed.

2. In your master class, subclass your FieldPut method with something like
this:

	METHOD FieldPut( uField, uValue ) CLASS YourMasterClass
	IF SELF:lTracking
		IF !( SELF:FieldGet( uField ) == uValue )	// There's a change
			SELF:cChanges += "your description for changes"
			SELF:lRecordChanged := TRUE			// flag it
		ENDIF
	ENDIF
	RETURN SUPER:FieldPut( uField, uValue )

3. Then use the Notify() method to detect when to save your audit info:

	METHOD Notify( kNotify, uDescription ) CLASS YourMasterClass
		LOCAL uReturn		AS USUAL

		// Save return value
		uReturn := SUPER:Notify( kNotify, uDescription )

		// You're interested in saving when trying to move
		IF SELF:lTracking .and. IF SELF:lRecordChanged
			// Save your data
			SELF:SaveAudit()
		ENDIF

		RETURN uReturn

4. Create your saving

	METHOD SaveAudit() CLASS YourMasterClass

		// Save data
		// ...
		// ...

		// Reset flags and variables
		SELF:lRecordChanged := FALSE
		SELF:cChanges := ""

		RETURN NIL

5. Subclass your Commit() to save when it is called directly

		METHOD Commit() CLASS YourMasterClass

			IF SELF:lTracking .and. SELF:lRecordChanged
				SELF:SaveAudit()
			ENDIF

			RETURN SUPER:Commit()

This is a basic idea but should put you on track with the subject.

HTH,
Amilcar A. Camargo F.
Guatemala, C. A.
http://acf1996.webatu.com/cavo

[toc] | [prev] | [next] | [standalone]


#1648

FromNick Friend <nicktekhne@googlemail.com>
Date2011-07-08 01:38 -0700
Message-ID<fb5f831b-ecae-485b-a7d6-b5c6416b758a@w4g2000yqm.googlegroups.com>
In reply to#1642
Thanks for the input Amilcar. I actually had a good bash at this last
night in a fit of inspiration and it's all working very well
apparently. I've taken a slightly different approach, pretty much what
I outlined in the first message, but the principle is the same and
I've just added auditing to our entire application with about 3-4
hours work in total.

Of course we'll have to see what the performance impact is, but the
code is very concise, probably no more than 4 or 5 lines in any given
subclassed method, and the only part that uses any serious resources
is writing the audit record to the audit log, so I think it should be
minimal. There would be an issue if the audit log was to be allowed to
accumulate, as it would grow very quickly as others pointed out, but
that's not the case for us.

Nick

[toc] | [prev] | [next] | [standalone]


#1651

Fromdlzc <dlzc1@cox.net>
Date2011-07-08 10:28 -0700
Message-ID<4a694cdf-53b6-4ef3-81c8-1bea11574cd2@t7g2000vbv.googlegroups.com>
In reply to#1648
Dear Nick Friend:

On Jul 8, 1:38 am, Nick Friend <nicktek...@googlemail.com> wrote:
> Thanks for the input Amilcar. I actually had a good
> bash at this last night in a fit of inspiration and
> it's all working very well apparently. I've taken a
> slightly different approach, pretty much what I
> outlined in the first message, but the principle is
> the same and I've just added auditing to our entire
> application with about 3-4 hours work in total.

Can you drop a better hint as to how you instruct DbServer to audit or
not?

Do you use a different USE call for the files (meaning future program
mods need to know to do this), did you encapsulate opening those files
in their own routine with the necessary modifications, or did you
"teach" DbServer which files needed to be audited?

I don't know VO, and posterity may understand what you've said, but...

David A. Smith

[toc] | [prev] | [next] | [standalone]


#1683

From"richard.townsendrose" <richard.townsendrose@gmail.com>
Date2011-07-16 03:50 -0700
Message-ID<796f1763-6414-4c1e-9393-6d2b15094fb7@df3g2000vbb.googlegroups.com>
In reply to#1651
Nick

any chance that you could post your code ... or email it ...

the ads stuff was really useful ...

richard

[toc] | [prev] | [next] | [standalone]


#1692

FromNick Friend <nicktekhne@googlemail.com>
Date2011-07-18 00:24 -0700
Message-ID<8183d0b9-39b2-4c1f-9ed7-00f6465ee2f1@h12g2000vbx.googlegroups.com>
In reply to#1683
Sure... I've been refining and testing the code.... nearly done. In a
few days I'll send a copy to those that have requested it.

Nick

On Jul 16, 11:50 am, "richard.townsendrose"
<richard.townsendr...@gmail.com> wrote:
> Nick
>
> any chance that you could post your code ... or email it ...
>
> the ads stuff was really useful ...
>
> richard

[toc] | [prev] | [next] | [standalone]


#1696

FromBenito Fernández <bfernandezm@prodigy.net.mx>
Date2011-07-18 14:13 -0500
Message-ID<j020lg$hji$1@speranza.aioe.org>
In reply to#1692
Nick,

"Nick Friend"  escribió en el mensaje de 
noticias:8183d0b9-39b2-4c1f-9ed7-00f6465ee2f1@h12g2000vbx.googlegroups.com...

> Sure... I've been refining and testing the code.... nearly done. In a
> few days I'll send a copy to those that have requested it.
>
> Nick

I am interested too. One of my clients is requesting some sort of auditing.

TIA,
Benito Fernández

On Jul 16, 11:50 am, "richard.townsendrose"
<richard.townsendr...@gmail.com> wrote:
> Nick
>
> any chance that you could post your code ... or email it ...
>
> the ads stuff was really useful ...
>
> richard 

[toc] | [prev] | [next] | [standalone]


#1697

FromKarl Faller <k.faller_withoutthat_@onlinehome.de>
Date2011-07-18 21:17 +0200
Message-ID<0m1927hhf25rdb3eb7en5renmqemc2c94p@4ax.com>
In reply to#1692
Nick,
me too, please ;-)

TIA
Karl 

>Sure... I've been refining and testing the code.... nearly done. In a
>few days I'll send a copy to those that have requested it.
>
>Nick
>
>On Jul 16, 11:50 am, "richard.townsendrose"
><richard.townsendr...@gmail.com> wrote:
>> Nick
>>
>> any chance that you could post your code ... or email it ...
>>
>> the ads stuff was really useful ...
>>
>> richard

[toc] | [prev] | [next] | [standalone]


#1716

Fromdengler <dewo001@hotmail.com>
Date2011-07-20 21:29 +0200
Message-ID<c2be27pug5pkom139v3f3pp647l2eb3n61@4ax.com>
In reply to#1692
Hi,
i am interested too
Thanks

Nick Friend <nicktekhne@googlemail.com> schrieb:

>Sure... I've been refining and testing the code.... nearly done. In a
>few days I'll send a copy to those that have requested it.
>
>Nick
>
>On Jul 16, 11:50 am, "richard.townsendrose"
><richard.townsendr...@gmail.com> wrote:
>> Nick
>>
>> any chance that you could post your code ... or email it ...
>>
>> the ads stuff was really useful ...
>>
>> richard

[toc] | [prev] | [next] | [standalone]


#1717

FromNick Friend <nicktekhne@googlemail.com>
Date2011-07-20 13:38 -0700
Message-ID<d38214a4-c446-43ac-874a-39118adc3069@h17g2000yqn.googlegroups.com>
In reply to#1716
Hi all,

Can anybody who's interested in getting this please drop me an email
so I can pick up your email addresses.... I've been horribly busy but
I'm hoping that by the beginning of next week I can take a bit of time
to organise this and send it out.

Anyway, it seems to be working perfectly (though of course no
guarantees ;-) ), and the only code change is to inherit from the sub-
classed DbServer. It's also very easy to filter what you want to
record and to change the storage mechanism for the audit.

Nick

[toc] | [prev] | [standalone]


Back to top | Article view | comp.lang.clipper.visual-objects


csiph-web