ADS
Marc Verkade
I had a discussion with R&D of ADS and they have seen and accepted the issue
with the OrderKeyNo and OrderKeyCount speed. See the messages below... The
only thing is that it wil take quite a bit of work to really solve the
issue.. How many of you are using ADS and want the OrderKeyNo and
OrderKeyCount issues solved??? Please drop a note at NG
devzone.advantagedatabase.com in the group Advantager.Beta.7.0
p.s. The supplied RDD does fix the OrderKeyCount issue, and gives a speed
increment of the OrderKeyNo issue, but is not as fast as the DBFCDX one...
So for better performance, ADS7.0 is the way to go.
Grtz, Marc
<START ADS TEXT 1>
Marc & others,
The RDD I posted will only give OrderKeyNo performance gains with fully
optimized filters. Here's a build that should perform better with any (or
no) filters. This is most likely the RDD we'll ship for 7.0. Give it a try
with AX_SetExactKeyPos( .F. ).
Like I said earlier, we've seen this issue before with our Delphi client and
giving a approximation for the key number was an acceptable solution. To
give exact values with performance similar to the VO RDD would require more
effort than possible at this time. With enough customer motivation it might
become a priority task for a future release, but that is management's
decision, not mine.
Let me know if this RDD helps.
Peter
</START ADS TEXT 1>
<START ADS TEXT 2>
7.0 beta VO users,
I've attached a modified version of our 7.0 beta VO RDD and DBFAXS library
for you to try. It includes OrderKeyCount and OrderKeyNo improvements. If
these modifications prove useful, they'll be included in the 7.0 release.
I'd like to improve the performance of OrderKeyNo even more, but that will
have to wait for our next release. Here's a description of the changes:
OrderKeyCount: For optimized filters, the entire DBF is no longer scanned.
The RDD (actually ACE) simply returns the AOF bit count. No changes to your
application are necessary.
OrderKeyNo: A new function has been added to the DBFAXS library called
AX_SetExactKeyPos. This function will toggle some internal logic to whether
you get an exact or estimated key number from OrderKeyNo. Calling
AX_SetExactKeyPos( .F. ) will change the behavior of OrderKeyNo to return an
approximate key number and improve its performance. The default is TRUE, to
simulate our original behavior.
Let me know how it works, all feedback is welcome :-)
--
Peter Funk
Advantage R&D
</START ADS TEXT 2>
Paul De Bie
I had emailed them a this issue one month ago too... It's the first thing
you'll bump into when you start using ADS I guess.
BTW; I'm having another issue with the optimized filters, which I mailed
them too, described in real detail. The answer I got was besides the
question/problem. Meanwhile, no further news. If you like; I can email you
this problem, maybe you can comment on it (since I recall you once said
somewhere that you had tested and benchmarked the ADS optimized filters...)?
NB: dropped the note at the ADS group...
Paul
"Marc Verkade" <ma...@caiw.nl> schreef in bericht
news:3f4bef65$0$49111$e4fe...@news.xs4all.nl...
Nathan Robeson
I too have encountered this issue before and have talked to Extended Systems
about it. It becomes impossible on huge tables; not only does it take a long
time, but while it builds a AOF, the CPU in the server is eaten up and the
other clients must wait. I had to work around this issue Glad to hear 7.0
addresses it.
Just want to say however that ADS is a super product and allows us to do
things that could never be done with native drivers.
Nathan
"Marc Verkade" <ma...@caiw.nl> wrote in message
news:3f4bef65$0$49111$e4fe...@news.xs4all.nl...
Marc Verkade
Drop a note at the ADS NG.. It will help..
I asked my Dutch rep to climb the ladder internally to have the issues
adressed..
When triing both ways, I hope the issues will be solved soon.
GRtz, Marc
"Nathan Robeson" <nat...@voicecue.com> schreef in bericht
news:vkpocjl...@news.supernews.com...
Marc Verkade
news://devzone.advantagedatabase.com
grtz, Marc
"Marc Verkade" <ma...@caiw.nl> schreef in bericht
news:3f4bef65$0$49111$e4fe...@news.xs4all.nl...
ilias
wrote:
>Hello,
>I had a discussion with R&D of ADS and they have seen and accepted the issue
>with the OrderKeyNo and OrderKeyCount speed. See the messages below... The
>only thing is that it wil take quite a bit of work to really solve the
>issue.. How many of you are using ADS and want the OrderKeyNo and
>OrderKeyCount issues solved??? Please drop a note at NG
>devzone.advantagedatabase.com in the group Advantager.Beta.7.0
>
>p.s. The supplied RDD does fix the OrderKeyCount issue, and gives a speed
>increment of the OrderKeyNo issue, but is not as fast as the DBFCDX one...
Do i understand correct?:
the ADS7.0 is not as fas as the DBFCDX (in this issue).
[...]
><START ADS TEXT 1>
>Marc & others,
>The RDD I posted will only give OrderKeyNo performance gains with fully
>Let me know how it works, all feedback is welcome :-)
[...]
where is the exact location of thos ADS messages?
(news-server, news-group, message-subject)
Jean-Paul Bleau
Jean-Paul Bleau
Geoff Schaller
"Jean-Paul Bleau" <BleauJP***REMOVE***@Hotmail.com> wrote in message
news:binfve$949j5$1...@ID-144863.news.uni-berlin.de...
Greg Garza
Unfortunately for idios the ADS NGs are moderated. And if he was paying
attention he would have the information he's asking for.
Greg
ilias
>On Wed, 27 Aug 2003 01:31:35 +0200, "Marc Verkade" <ma...@caiw.nl>
>wrote:
>
>>Hello,
[...]
>>p.s. The supplied RDD does fix the OrderKeyCount issue, and gives a speed
>>increment of the OrderKeyNo issue, but is not as fast as the DBFCDX one...
>
>Do i understand correct?:
>
>the ADS7.0 is not as fas as the DBFCDX (in this issue).
yes.
>
>[...]
>><START ADS TEXT 1>
>>Marc & others,
>>The RDD I posted will only give OrderKeyNo performance gains with fully
>>Let me know how it works, all feedback is welcome :-)
>[...]
>
>where is the exact location of thos ADS messages?
>
>(news-server, news-group, message-subject)
news://devzone.advantagedatabase.com:119/Advantage.Beta.7.0
drop a note on this message:
news://devzone.advantagedatabase.com:119/3f4b...@solutions.advantagedatabase.com
-
ilias
wrote:
>Hello,
>I had a discussion with R&D of ADS and they have seen and accepted the issue
>with the OrderKeyNo and OrderKeyCount speed. See the messages below... The
[note: navigation information, mainly for usenet-archives]
see a related discussion:
[ADS] - Extended Systems Inc. does not care about VO!
http://groups.google.com/groups?threadm=dif1lvos2s8qnmj6v6khn0h4s1e8c6bb6d%404ax.com
Marc Verkade
Do not use my stuff and do not quote me please.
This is a testcase for ADS to let them see what is happening.
They looked at the file, agreed with the problem, and arranged a fix in 2
days.
They are fixing it completely in a next build.
No grtz, Marc
ps The file is deleted.
"ilias" <use...@abeon.com> schreef in bericht
news:vn60lv0caf3ivlm5o...@4ax.com...
ilias
wrote:
>Ilias,
>Do not use my stuff and do not quote me please.
Please don't write in public newsgroups, if you don't want to be
quoted.
Please don't post public links to stuff, if you don't want that stuff
to be used [e.g. by placing a link to it].
>This is a testcase for ADS to let them see what is happening.
So why don't let the VO-community benefit from your effort?
>They looked at the file, agreed with the problem, and arranged a fix in 2
>days.
For a problem that is commonly known since 2 years...
>They are fixing it completely in a next build.
Looks VO has no priority.
>No grtz, Marc
>
>ps The file is deleted.
>> see: www.marti.nl/ADSTest.zip
I don't even downloaded it.
You should be so friendly to republish it, thus other users that read
advantage news or e.g. google archives can recreate the case and check
easily if the problem is solved.
Think about it.
If links published in usenet don't persist, usefulness of this media
decreases.
You don't do a favour to me, but to the current and future
VO-community.