Interbase 6.5 released!
Ben Matterson
"Borland InterBase 6.5 is a high-performance, cross-platform, SQL
standards-compliant relational database that combines ease of use, low
maintenance costs, and enterprise power. Certified by Borland, InterBase
6.5 provides the flexibility required for Internet, mobile, and embedded
database applications. InterBase supports all popular desktop clients
and application-builder frameworks."
"High-Performance Borland InterBase 6.5 Features Cross-Platform Support
and Enhanced Security for Developing and Deploying Internet, Mobile, and
Embedded Distributed Database Applications"
Ben
--
InterBase Technical Support
Borland Software Corp.
http://community.borland.com/search/ib_knowledgebase.html
Martijn Tonies
--
Martijn Tonies
InterBase Workbench - the developer tool for InterBase and Firebird
http://www.interbaseworkbench.com
Upscene Productions
http://www.upscene.com
"This is an object-oriented system.
If we change anything, the users object."
"Ben Matterson" <bmatt...@borland.com> wrote in message
news:3C0D0EE1...@borland.com...
Aaron Ruddick[Borland]
We don't have any plans for an IB6.5 trial at this time. You can get the IB65
desktop download from:
http://shop.borland.com/FrontDoor/0,1076,3-15,00.html
Or the free IB6.0 trial: http://www.borland.com/interbase/downloads/
--
Thank you,
Aaron Ruddick
InterBase QA
http://www.borland.com/interbase
Andy Murphy
Let me put it this way, is it any quicker?
Andy.
Dan Palley
There were improvements made to the over-the-wire protocol that should make
a difference over slower connections (i.e. VARCHAR's are no longer padded
out to their full length).
Is there any particular feature that seems to need speed improvement (other
than the optimizer itself)?
Dan
"Andy Murphy" <an...@rating-software-company.co.uk> wrote in message
news:3c0d2c3c_2@dnews...
Andy Murphy
near this whilst compiling our lists and avoid it by backing up and
restoring (which takes about 3-4 hours) or the ability as I have seen
described in the help files (but it doesnt work) to declare to an sql
statemet or stored proc 2 databases and then do cross database queries i.e.
Select * from db1.table1 where etc etc
OK rant over... <g>
David R. Robinson
Quoting Jon Arthur from:
http://community.borland.com/article/0,1410,28083,00.html
"...Plus, InterBase 6.5 offers a host of new features, such as Enhanced
Metadata security, VLDB Support (64 bit I/O), Asynchronous Statement
Cancel and XML Data-generation."
David R.
Andy Murphy
(other
> than the optimizer itself)?
Well I realise i'm a bit of a different case because we use IB to power
large read-only database information systems. But we are also developing a
read/write multi-user, multi-tier management system which will plug into our
read only DB's so Interbase was chosen with this in mind.
So improvments, a faster record count, would be ideal, also we have to
compile over 6million records and we end up with 4.2 million after lots of
cross referencing. It takes five days to do this. We use a P4 2.0 ghz with a
4 channel raid card (in mode 0) with 4 x 15k rpm Ultra 160+ SCSI drives and
2gb of Rimm 800 mhz memory and its still slow. If we could use multiple /
quad processors it would make a lot of difference. Im thinking of putting
Linux on this PC instead of Win NT5 to see what difference it makes.
Whilst we are on improvements, what about a new Console that is a bit more
user friendly, i.e. graphic. and more up to date?
David R. Robinson
Dan Palley
will assist
> developers with the opensource version. New features found in InterBase
6.5 will
> be available only with the certified version.
This is unfortunate. Sounds like the Open Source version is pretty much a
dead-end.
"David R. Robinson" <ibin...@mail.com> wrote in message
news:3c0d3bcb_1@dnews...
> http://community.borland.com/article/0,1410,28083,00.html
>
>
>
Dan Palley
without having to back up and restore. 6.5 allows a single file to be
bigger than 4 GB.
The cross database query technique you mention only works when using the
BDE; it's not a function of Interbase. There are other commercial products
that let you do the same thing (Easysoft has one).
Dan
"Andy Murphy" <an...@rating-software-company.co.uk> wrote in message
news:3c0d3916$1_2@dnews...
J Hall
wrote...
Hi,
> For more information go to http://www.borland.com/interbase.
Is there a Changelog of the fixes? (the Sourceforge tree does not seam
to have been updated yet)
As the new features are a bit "thin", I would guess that most of the
past year has been spent fixing bugs and other minor improvements, which
is currently the biggest difference between IB and Firebird.
I am personally a bit disappointed that features that are 'identical'
between IB and Firebird have been implemented with a different syntax,
but I suppose we will all have to make the "choice" sooner or later.
J
Craig Stuntz (TeamB)
Andy Murphy wrote:
>
> Let me put it this way, is it any quicker?
One other point here: IB 6.5 uses memory in a more efficient way (but
don't ask me for details) so it should make better use of available
resources. Also, the correlation between increasing buffers and
increasing restore time (as in Rob's test) has been fixed. So there are
speed improvements separate from the optimizer, CPU use, and the other
things which folks have mentioned.
-Craig
--
Craig Stuntz (TeamB) · Vertex Systems Corp. · Columbus, OH
We're hiring: http://www.vertexsoftware.com/careerops.htm#sd
Delphi/InterBase WebLog: http://delphi.weblogs.com
Craig Stuntz (TeamB)
Andy Murphy wrote:
>
> So improvments, a faster record count, would be ideal, also we have to
> compile over 6million records and we end up with 4.2 million after lots of
> cross referencing.
Honestly, my recommendation for record counting is don't do it.
There's no good way to get a record count in a multiuser, versioning
environment which is both accurate and fast. *Most* uses of record
counting can be replaced with other tests (e.g., EXISTS) and those which
can't should be kept to a minimum and heavily optimized.
> It takes five days to do this. We use a P4 2.0 ghz with a
> 4 channel raid card (in mode 0) with 4 x 15k rpm Ultra 160+ SCSI drives and
> 2gb of Rimm 800 mhz memory and its still slow. If we could use multiple /
> quad processors it would make a lot of difference.
In my experience, the biggest bottleneck is your HDD throughput, so
before looking for help from multiple processors I'd try a caching RAID
configuration.
> Whilst we are on improvements, what about a new Console that is a bit more
> user friendly, i.e. graphic. and more up to date?
AFAIK, Borland still plans to release IBConsole updates separately --
i.e., it will support 6.5 but you won't need to have 6.5 to get the
IBConsole updates.
Craig Stuntz (TeamB)
Dan Palley wrote:
>
> > We will continue to update the opensource tree with many bug fixes that
> will assist
> > developers with the opensource version. New features found in InterBase
> 6.5 will
> > be available only with the certified version.
>
> This is unfortunate. Sounds like the Open Source version is pretty much a
> dead-end.
Jon announced this summer that the Open Edition will get all of the bug
fixes and he's sticking to his word, so I don't think it's fair to call
it a "dead end." The IB team has chosen not to add any new features to
the Open Edition this time around -- I would speculate that it's because
they want to give people a greater reason to buy the certified
editions. But that doesn't mean they won't be added in the future.
Some people have proposed that features would be added to OE one version
behind the certified editions, for example. I don't know if this is
what the IB team will choose, but I would have been very surprised to
see the features from 6.5 go into OE, because product differentiation
(certified vs. OE) has been a problem since the certified release.
Jason Wharton
make
> a difference over slower connections (i.e. VARCHAR's are no longer padded
> out to their full length).
It is VERY nice to know that my complaint of this was heard (possibly)!!
I remember getting beat up pretty hard for suggesting whitespace be
suppressed prior to feeding into the output data stream. A certain vocal
person who shall remain unnamed said that was foolish since RLL compression
was used and all whitespace would be eliminated anyway.
The point I labored hard to make, and felt that I had failed and decided to
give up on, was that records are fed into a fixed buffer frame and then
compressed. The sooner the buffer frame filled meant the fewer records were
nestled together in a single network packet, which to me signified a less
efficient wire protocol.
What I wanted to see happen was the white space suppressed prior to
inserting into the buffer frame. This would allow many more records into it
and then compression could be applied as appropriate for the data itself.
Density, rather than just compression was what I was addressing.
The one point I wasn't sure of was if the records were RLL compressed prior
to placing them in the buffer frame or not. But, where I knew I really had a
point was inside of the VARCHAR whitespace (meaning the bytes beyond the
length indicator for the significant characters) whatever junk bytes I sent
up the wire and selected came back.
This not only told me that the junk data was eating bandwidth on the wire
but it was also eating space inside of my GDB file.
(Keep in mind, I've been using IB for a very long time and some of these
issues may have already been resolved in IB 6.)
So, hearing this is very encouraging that they have not only cleaned up the
wire protocol but the internal record steam storage management. If nothing
else, zapping whitespace to null chars so they compress better under the
current conditions (to keep a consistent ODS) is fine too.
Nice work gentleman.
FWIW,
Jason Wharton
CPS - Mesa AZ
http://www.ibobjects.com
Jason Wharton
> large read-only database information systems. But we are also developing a
> read/write multi-user, multi-tier management system which will plug into
our
> read only DB's so Interbase was chosen with this in mind.
>
> So improvments, a faster record count, would be ideal, also we have to
> compile over 6million records and we end up with 4.2 million after lots of
> cross referencing. It takes five days to do this. We use a P4 2.0 ghz with
a
> 4 channel raid card (in mode 0) with 4 x 15k rpm Ultra 160+ SCSI drives
and
> 2gb of Rimm 800 mhz memory and its still slow. If we could use multiple /
> quad processors it would make a lot of difference. Im thinking of putting
> Linux on this PC instead of Win NT5 to see what difference it makes.
One item you may not be aware of that will help cut down the I/O when
generating your RO database is to make use of the "fill factor" setting. I
believe it is also referred to as the "no reserve" setting. What it is
controls is the behavior of leaving spare room on pages or not when
inserting. In short, if you are anticipating records will be either updated
or deleted it is much faster if there is a little bit of room on each page
to store the update/delete without having to span to a new page. This
especially helps because record chains often need to be walked and it is a
plus when they are located on the same page. This option is on by default.
But, if you are not anticipating updates or deletes to records (your case),
you should set the "no reserve" option on so that you will get a complete
page fill when doing inserts. This will make your records more dense on the
pages, use less disk space, and take 20% less I/O to generate the GDB. It
will also have fewer pages to walk when doing record counts.
In short, how does a 20% discount sound?
Performance is on sale today. <g>
> Whilst we are on improvements, what about a new Console that is a bit more
> user friendly, i.e. graphic. and more up to date?
If you want a nice little console tool, take a peek at IB_SQL. It is an ISQL
replacement tool which once you get the hang of, becomes indispensable. It's
designed not to take over your desktop and is loaded with tons of
functionality. It is my freeware contribution to the community. It has a
help file too.
http://www.ibobjects.com/ibo_IB_SQL.html
Rumor has it, IB_SQL is going to be chosen as the default/official ISQL tool
for Firebird when I finish porting it to LINUX. Once that happens, I'm sure
it will enjoy many more continued improvements. I'll even add the "no
reserve" switch on there if you like. <g>
Regards,
Craig Stuntz (TeamB)
J Hall wrote:
>
> Is there a Changelog of the fixes? (the Sourceforge tree does not seam
> to have been updated yet)
Not yet, but the press release did announce that SourceForge would be
updated, so I would imagine it's coming. That won't be the full list of
bug fixes, though, since there are probably some fixes which only
pertain to the certified editions (anything to do with licensing, for
example).
> As the new features are a bit "thin", I would guess that most of the
> past year has been spent fixing bugs and other minor improvements, which
> is currently the biggest difference between IB and Firebird.
I would guess they also spent some time laying the groundwork for
features which will be in 7, as well. For example, it's not
unreasonable to presume that the ability to cancel a query might help
with implementing SMP support (think about it a little). Another
difference between IB and Firebird which takes a lot of time is the full
test suite which each release of IB has to pass.
> I am personally a bit disappointed that features that are 'identical'
> between IB and Firebird have been implemented with a different syntax,
Which features are those? Firebird doesn't have asynchronous query
cancellation, XML support, or 6.5's metadata security, AFAIK. VLDB (>4
GB) doesn't have a syntax, and IB's ROWS feature has PERCENT and WITH
TIES options which Firebird's syntax wouldn't support, also AFAIK. IB
needs an ODS bump for the new features, so the databases produced by IB
and Firebird may be binary-incompatible.
FWIW, I also prefer the ROWS syntax because of its greater flexibility
with a UNION query.
Dan Palley
into the Open Edition). At the moment, there's nothing earth-shattering
(other than the ability to cancel a query) that would motivate my users to
buy 6.5; when 7.0 rolls out, that's another story, though.
Dan
"Craig Stuntz (TeamB)" <cstuntz@no_spam.vertexsoftware.com> wrote in message
news:3C0D4D26.458F947B@no_spam.vertexsoftware.com...
J Hall
cstuntz@no_spam.vertexsoftware.com wrote...
Hi,
> > I am personally a bit disappointed that features that are 'identical'
> > between IB and Firebird have been implemented with a different syntax,
>
> Which features are those?
I don't want to start anything here, especially as we only have the 'new
features.pdf' to go on.
> Firebird doesn't have asynchronous query cancellation
Agreed
> XML support
A few hours word <g>
> or 6.5's metadata security, AFAIK.
Agreed.
> VLDB (>4 GB) doesn't have a syntax,
Agreed.
> and IB's ROWS feature has PERCENT and WITH
> TIES options which Firebird's syntax wouldn't support
You obviously know more than we do on this feature, so I shall wait till
the SF tree is updated. But when I read that, I immediately thought of
LIMIT/FIRST/etc
> also AFAIK. IB
> needs an ODS bump for the new features, so the databases produced by IB
> and Firebird may be binary-incompatible.
Which is something the FB people having been holding back on.
> FWIW, I also prefer the ROWS syntax because of its greater flexibility
> with a UNION query.
I shall wait till the SF update to make a comment on this.
J
J Hall
cstuntz@no_spam.vertexsoftware.com wrote...
Hi,
> The IB team has chosen not to add any new features to
> the Open Edition this time around.
Have they? Thats disappointing.
> I would speculate that it's because
> they want to give people a greater reason to buy the certified
> editions.
I am sure we would all buy it, if there was a build of the 'open
edition' without the per server/user licensing.
> I would have been very surprised to
> see the features from 6.5 go into OE, because product differentiation
> (certified vs. OE) has been a problem since the certified release.
There should not be a difference IMHO, apart from the building/testing
and 'value add features'.
j
Andy Murphy
As far as record counting is concerned, if you supply a large list to
someone for say marketing purposes they might want to look at various
criteria to see if the number of records returned from the query i.e.
'prospective cutomers' is worth further investigation.
I supply a list of information for every commercial property in the UK. To
any one of our clients, how many 'shops' in a certain area over a cetain
value that havent changed size in 12 years, for example, could be very
important. In some cases the record count is more important than the data,
strange but true.
To be honest its not that bad, but not as good as typical flat file
databases. I know, chalk and cheese <g>
Andy Murphy
On the point of 'Use All Space' , all my customers get a database that is
restored using all space.
I think I confused my point, from my customers point of view the only thing
that would be nice is the record counting a bit quicker. The App/Database
runs fine on a PII 300 laptop!
From my point of view before they get this 100% filled nice tidy database,
some mug (err me) has to compile a huge amount of data and that is done with
80% fill, thats where i need the extra power of multi-processors or as Craig
points out faster HDD subsystems would help out.
Anyway food for thought.
Thanks.
Bill Todd (TeamB)
The cache manager has been fixed so that caches > 10k pages will not
decrease performance. The cache manager has also been improved to provided
faster cache allocation and less cache management overhead so using a larger
cache may improve your performance. As Dan said, SMP is an IB 7 feature.
--
Bill
(TeamB cannot answer questions received via email)
Wayne Niddery [TeamB]
> Does it have multi processor support yet?
Not yet (slated for version 7), but as an interim improvement, you can now
tell IB which processor to run on (affinity), this stops the performance
degradation on multi-processor Windows boxes.
--
Wayne Niddery (Logic Fundamentals, Inc.)
RADBooks: http://www.logicfundamentals.com/RADBooks/delphibooks.html
Try to make original mistakes, rather than needlessly repeating [others]. -
Donald Rumsfeld, U.S. Secretary of Defense
Jeff Overcash (TeamB)
"Wayne Niddery [TeamB]" wrote:
>
> "Andy Murphy" <an...@rating-software-company.co.uk> wrote in message
> news:3c0d2c3c_2@dnews...
> > Does it have multi processor support yet?
>
> Not yet (slated for version 7), but as an interim improvement, you can now
> tell IB which processor to run on (affinity), this stops the performance
> degradation on multi-processor Windows boxes.
>
It also defaults to the 1st CPU if you don't add the config parameter to tell it
otherwise.
--
Jeff Overcash (TeamB)
(Please do not email me directly unless asked. Thank You)
The fool escaped from paradise will look over his shoulder and cry
Sit and chew on daffodils and struggle to answer why?
As you grow up and leave the playground
Where you kissed your Prince and found your frog
Remember the jester that showed you tears, the script for tears. (Fish)
Andres Allkivi
data and app it may be significantly faster or not... you have to try.
Andres
Claudio Valderrama C.
>
> VLDB (>4
> GB) doesn't have a syntax
What's a syntax for VLDB?
C.
Andy Murphy
> IB 6.5 includes 64 bit file i/o so the 2/4 gigabyte file size limit is
gone.
> The cache manager has been fixed so that caches > 10k pages will not
> decrease performance. The cache manager has also been improved to provided
> faster cache allocation and less cache management overhead so using a
larger
> cache may improve your performance. As Dan said, SMP is an IB 7 feature.
This sounds good. You see I'm quite happy for my clients to run opensource
or firebird, but to compile the data, I'm quite prepared to go and buy 6.5
server plus say 1-3 licences to do the work on, if its significantly
faster/better. If not I will wait for 7 to come along.
Andy.
Craig Stuntz (TeamB)
Andy Murphy wrote:
>
> I supply a list of information for every commercial property in the UK. To
> any one of our clients, how many 'shops' in a certain area over a cetain
> value that havent changed size in 12 years, for example, could be very
> important. In some cases the record count is more important than the data,
> strange but true.
See if there is a way you can reduce the number of times you count
records -- perhaps you can do it once and store the result for later
use.
HTH,
Craig Stuntz (TeamB)
J Hall wrote:
>
> > I would have been very surprised to
> > see the features from 6.5 go into OE, because product differentiation
> > (certified vs. OE) has been a problem since the certified release.
>
> There should not be a difference IMHO, apart from the building/testing
> and 'value add features'.
IMHO, they should do whatever they need to do to encourage purchase of
the certified editions, because that (+ support) is what funds IB's
development.
Craig Stuntz (TeamB)
I don't think there is such a thing (a syntax). It doesn't make sense.
That's why I didn't understand J Hall's comment about "identical
features with different syntax."
J Hall
cstuntz@no_spam.vertexsoftware.com wrote...
Hi,
> That's why I didn't understand J Hall's comment about "identical
> features with different syntax."
I was talking about the FIRST/TOP/LIMIT/Whatever syntax.
J
Aaron Ruddick[Borland]
We didn't change the 6.0 book set. So all the new features are covered in
the release notes on the cd (or Desktop download).
You have probably already looked at the brief description on our web site,
but here is the URL:
http://www.borland.com/about/press/2001/interbase65_available.html
--
Thank you,
Aaron Ruddick
InterBase QA
http://www.borland.com/interbase
Max
> updated, so I would imagine it's coming. That won't be the full list of
> bug fixes, though, since there are probably some fixes which only
> pertain to the certified editions (anything to do with licensing, for
> example).
Actually, the list of fixed bugs would be the most interesting thing to read
about 6.5. We all know about dozens of outstanding IB6 bugs. All core
bugfixes should (re. Jon Arthur) appear in IB OE at SourceForge, and I'm
(not alone) eager to see them. It will help us measure two things:
1) How much is Borland serious about IB OE. We know that Borland has an
antagonistic stance to IB OE for obvious reasons, but we still don't know
how much and to what direction they shift the balance with 6.5 release.
2) How many bugs they squashed (did they really fixed anything or just add
new features instead?) and how. And also how much they harvested the
Firebird project for code and ideas (I'm not against that on principle!). I
know that they monitor all firebird-devel traffic, and it would be
interesting to know if they learned something here :-)
> I would guess they also spent some time laying the groundwork for
> features which will be in 7, as well. For example, it's not
> unreasonable to presume that the ability to cancel a query might help
> with implementing SMP support (think about it a little). Another
> difference between IB and Firebird which takes a lot of time is the full
> test suite which each release of IB has to pass.
There is no other difference in FB and IB testing other than a method. And
to be honest, Firebird's testing phase is actually far longer and realized
at bigger scale than at Borland.
> Which features are those? Firebird doesn't have asynchronous query
> cancellation,
Ahhh! This heature is already in InterBase engine for long time. It's only
not active and tested. What a dangerous endeavour for Borland to provide it
in 6.5!!! And after all, I haven't had a single need for this feature in
last seven years I work with IB. But yes, it's nice to have a life-belt even
if one can swim.
> XML support,
Buzz-feature. Nothing more and nothing less. It would be better to implement
it in a half-day at IBX or IBO level for anyone who _really_ want it (I'm
surprised that there is not a Delphi component for that already available
for free at Torry. It looks that no one had the need for that before :). And
what IB provides is not even near to what other servers offer as "XML
support".
> or 6.5's metadata security,
That's nice feature, but it not really save a day for anyone concerned about
security. Yep, IB security is less flaky and broken with this, but it's like
to add a water-pump to already sunken ship.
> AFAIK. VLDB (>4 GB) doesn't have a syntax,
Where is the 64-bit I/O support for Linux ? Any plan to provide it in future
? From coder's perspective, it's not different from providing it on Solaris,
so why not Linux ? A-ha! Borland doesn't want to bite the bullet and provide
two versions for Linux !!! (for those who doesn't know, 64-bit I/O support
on Linux is kernel-version dependent).
> and IB's ROWS feature has PERCENT and WITH
> TIES options which Firebird's syntax wouldn't support, also AFAIK.
But FAIK Firebird's syntax is according to SQL standard, while Borland's is
not.
> IB
> needs an ODS bump for the new features, so the databases produced by IB
> and Firebird may be binary-incompatible.
Yep. It was expected. Borland needs to differentiate from other IB
incarnations as much as possible. Even minor change in ODS and over-wire
protocol will help Borland to break from drop-in-replacement gauntlet. It's
smart business decision, but I wory that it will not work for long (only
until solution would be provided). I expect more drastic ODS changes in IB
7.
> FWIW, I also prefer the ROWS syntax because of its greater flexibility
> with a UNION query.
That's what I like about it too. But this feature already attracts only few
developers (especially the percent and with ties variant), so it actually
doesn't count much.
Nice try Borland, but there is nothing among new features that would draw
$50 for Media Kit from my pocket, even if I'd sum them up. But fixed bugs
could be worth of $100, so I'll wait to see what's fixed.
SVM
David R. Robinson
resist this one. :)
> There is no other difference in FB and IB testing other than a method.
And
> to be honest, Firebird's testing phase is actually far longer and
realized
> at bigger scale than at Borland.
What proof do you have of this? Do you work at Borland and know how
they do their testing? They have an entirely new QA group and unless
you work at Borland it is highly unlikely that you know anything at all
about their QA processes at all. You have absolutely no basis for
making statements about Borland's testing.
> Ahhh! This heature is already in InterBase engine for long time. It's
only
> not active and tested. What a dangerous endeavour for Borland to
provide it
> in 6.5!!! And after all, I haven't had a single need for this feature
in
> last seven years I work with IB. But yes, it's nice to have a
life-belt even
> if one can swim.
Hogwash. Even if there is code that is ifdef'd out that was written 5
years ago to support cancelling queries, it is totally irrelevant. It's
not a feature until it has been activated in the shipping product. Your
"what a dangerous endeavor" statement is ridiculous. How can you say
that implementing a new feature is dangerous? Just because you don't
have a need for it doesn't mean that nobody else does.
> > XML support,
> Buzz-feature.
Well, duhhhh.... XML is a huge buzzword right now. They would be
stupid to not add something XML related to their marketing efforts.
David R.
Jeff Overcash (TeamB)
Max wrote:
>
> And also how much they harvested the
> Firebird project for code and ideas (I'm not against that on principle!). I
> know that they monitor all firebird-devel traffic, and it would be
> interesting to know if they learned something here :-)
>
Just an FYI, you don't know that. As a matter of fact I know that they did not
harvest anything from FireBird and that they were adamant to myself and Rob that
they would not even look at FireBird code. They have not been tracking FireBird
at all. I could not even convince them that they at least needed to look at the
buglist that FireBird has addressed even if the source is never looked at.
J Hall
Hi,
> As a matter of fact I know that they did not
> harvest anything from FireBird and that they were adamant to myself and Rob that
> they would not even look at FireBird code.
Which has meant they have wasted time and money fixing bugs that have
probably already been fixed, when they could have been spending it on
the ones that haven't.
> They have not been tracking FireBird
> at all. I could not even convince them that they at least needed to look at the
> buglist that FireBird has addressed even if the source is never looked at.
I would say they would be pretty stupid not to.
This is one area where I think there should be some communication
between the two parties, there are enough outstanding bugs to go round.
J
Max
"David R. Robinson" <ibin...@mail.com> wrote in message
news:3c0e4b06$1_1@dnews...
> I am trying very hard to stay out of these threads, but I couldn't
> resist this one. :)
neither I :)
> > There is no other difference in FB and IB testing other than a method.
> And
> > to be honest, Firebird's testing phase is actually far longer and
> realized
> > at bigger scale than at Borland.
>
> What proof do you have of this? Do you work at Borland and know how
> they do their testing? They have an entirely new QA group and unless
> you work at Borland it is highly unlikely that you know anything at all
> about their QA processes at all. You have absolutely no basis for
> making statements about Borland's testing.
How can one solve the Catch 22 ? Believe me, I can't even tell you that I
know what I know or did or did not :) But even if I'd put my (secret)
argument aside, just compare the amount of betatesters at FB and in
Borland's QA departments (that's about scale). FB is tested thoroughly for a
year, while AFAIK Borland as any other commercial software company do
testing at the end, so I can't be very far from truth that they tested it
for cca last 3 months (at best).
> > Ahhh! This heature is already in InterBase engine for long time. It's
> only
> > not active and tested. What a dangerous endeavour for Borland to
> provide it
> > in 6.5!!! And after all, I haven't had a single need for this feature
> in
> > last seven years I work with IB. But yes, it's nice to have a
> life-belt even
> > if one can swim.
>
> Hogwash. Even if there is code that is ifdef'd out that was written 5
> years ago to support cancelling queries, it is totally irrelevant. It's
> not a feature until it has been activated in the shipping product. Your
> "what a dangerous endeavor" statement is ridiculous. How can you say
> that implementing a new feature is dangerous? Just because you don't
> have a need for it doesn't mean that nobody else does.
You're right, I wanted to point out that Borland just took a thing that was
partialy implemented and could be finished and tested in short time. What
they did in last year and half ? Just few little things, some already
partially implemented in IB6 (except metadata security that is also
disputable). Well, it's a minor version release, but anyway. I would be more
pleased if they'd pick and finish the expression indices.
Regards
Jeff Overcash (TeamB)
J Hall wrote:
>
> In article <3C0E4BB8...@onramp.net>, over...@onramp.net wrote...
>
> Hi,
>
> > As a matter of fact I know that they did not
> > harvest anything from FireBird and that they were adamant to myself and Rob that
> > they would not even look at FireBird code.
>
> Which has meant they have wasted time and money fixing bugs that have
> probably already been fixed, when they could have been spending it on
> the ones that haven't.
>
They can't just take FireBirds code and put it into IB . They have to approach
the copyright owner of that code and ask them permission to put it in. It has
been very clear that many of the FB people would never have given such
permission with the general hostility that has come from some of the FB people.
There has not been a waste of time here AFAICT.
> > They have not been tracking FireBird
> > at all. I could not even convince them that they at least needed to look at the
> > buglist that FireBird has addressed even if the source is never looked at.
>
> I would say they would be pretty stupid not to.
>
I agree, otherwise I would not have spent the last 3-4 months telling them they
need to look at the FB buglist. I understand their need to stay sandboxed from
FB though. I have the same problem with IBX.
> This is one area where I think there should be some communication
> between the two parties, there are enough outstanding bugs to go round.
>
I do not expect communications to ever start again. I have read some of the
communications (names removed) that have come to the FB camp to Borland. I
wouldn't work with them either if that was how I was going to get treated. This
is not to been seen as a blanket statement about even a majority of the
developers in FB, but enough are so anti Borland and bash them at any occasion
that I highly doubt Borland will work with them as long as these people are
still around. All IMO of course.
> J
Max
"Jeff Overcash (TeamB)" <over...@onramp.net> wrote
>
>
> Just an FYI, you don't know that. As a matter of fact I know that they
did not
> harvest anything from FireBird and that they were adamant to myself and
Rob that
> they would not even look at FireBird code. They have not been tracking
FireBird
> at all. I could not even convince them that they at least needed to look
at the
> buglist that FireBird has addressed even if the source is never looked at.
I know.
ssh...@borland.com is subscribed to firebird-devel. Enough?
Jeff Overcash (TeamB)
I have no idea who that is. It is not an e-mail address from any one on the
R&D, QA, Dev Support or management with InterBase. So no it is not enough. I
personally have talked to all of the R&D engineers on this issue. I know what
they have told me and you have presented no evidence at all to the contrary.
Nothing was taken from the Firebird sources. Doing so without the copyright
holder's permission would open Borland up to lawsuits and they are not going to
be stupid enough to do that.
Craig Stuntz (TeamB)
Max wrote:
>
> How can one solve the Catch 22 ? Believe me, I can't even tell you that I
> know what I know or did or did not :) But even if I'd put my (secret)
> argument aside, just compare the amount of betatesters at FB and in
> Borland's QA departments (that's about scale). FB is tested thoroughly for a
> year, while AFAIK Borland as any other commercial software company do
> testing at the end, so I can't be very far from truth that they tested it
> for cca last 3 months (at best).
You don't understand how automated testing works? Take a look at the
InterBase Test Control Suites -- Borland released them to the public,
but they can't release the test data used for IB (because it contains
proprietary customer data). This release of InterBase has the
equivilent of 15 years of testing performed, not to mention the beta
cycle.
Craig Stuntz (TeamB)
J Hall wrote:
>
> I was talking about the FIRST/TOP/LIMIT/Whatever syntax.
...Which is not an identical feature. Similar, perhaps, but there's no
reason IB should constrain itself to MySQL syntax. The flavor IB uses
is much more flexible.
Sean Leyne
latest 6.5 release.[The discussion was been 'interesting' enough, so I
haven't had to chime in ;-)) ].
> Just an FYI, you don't know that. As a matter of fact I know that they
did not
> harvest anything from FireBird and that they were adamant to myself and
Rob that
> they would not even look at FireBird code. They have not been tracking
FireBird
> at all.
That's not true, I know for a FACT that a member of the R&D team is
subscribed to the Firebird-dev list. While this is not poaching any code,
the level of discussion on the list can be very very detailed/technical in
nature.
It's unfortunate that Borland would be releasing much to the OE version. So
we won't able able to 'settle' the issue of whether Borland did any code
poaching.
--
Sean Leyne
There is nothing wrong with Interbase,
that can't be fixed with Firebird.
http://FirebirdSQL.org
Craig Stuntz (TeamB)
Max wrote:
>
> > Which features are those? Firebird doesn't have asynchronous query
> > cancellation,
>
> Ahhh! This heature is already in InterBase engine for long time. It's only
> not active and tested. What a dangerous endeavour for Borland to provide it
> in 6.5!!!
Your description is wrong. Borland did not simply surface existing
functionality. It wouldn't have worked if they did.
> > XML support,
>
> Buzz-feature. Nothing more and nothing less.
I agree. That said, there's nothing particularly wrong with this.
> > or 6.5's metadata security,
>
> That's nice feature, but it not really save a day for anyone concerned about
> security. Yep, IB security is less flaky and broken with this, but it's like
> to add a water-pump to already sunken ship.
I don't agree. Yes, it's a first step, but it's a vital one.
> > AFAIK. VLDB (>4 GB) doesn't have a syntax,
>
> Where is the 64-bit I/O support for Linux ? Any plan to provide it in future
> ? From coder's perspective, it's not different from providing it on Solaris,
> so why not Linux ? A-ha! Borland doesn't want to bite the bullet and provide
> two versions for Linux !!! (for those who doesn't know, 64-bit I/O support
> on Linux is kernel-version dependent).
Your speculation is totally wrong. IB 6.5, as released yesterday,
provides VLDB support for Linux kernels which support it (i.e., 2.4), as
well as < 4GB DB support for kernels which don't, in a single version.
> > and IB's ROWS feature has PERCENT and WITH
> > TIES options which Firebird's syntax wouldn't support, also AFAIK.
>
> But FAIK Firebird's syntax is according to SQL standard, while Borland's is
> not.
No, it's according to MySQL "standard."
> Nice try Borland, but there is nothing among new features that would draw
> $50 for Media Kit from my pocket, even if I'd sum them up. But fixed bugs
> could be worth of $100, so I'll wait to see what's fixed.
Actually, I'm impressed that there is no charge for the update beyond
the media kit. I don't think Borland makes any money on the media kits,
so this release is something of a gift to customers.
David R. Robinson
commercial software company do testing at the end, so I can't be very
far from truth that they tested it for cca last 3 months (at best).
Ok, I'll buy that as a legitimate theory. However, if you consider that
the FB group is made up of people working on it part time when they have
time to work on it (in addition to their real jobs), I'm not sure that
the 12 months is any different than your claim of 3 months that Borland
does. Either way, we're splitting hairs so no sense debating it.
>> You're right, I wanted to point out that Borland just took a thing
that was partialy implemented and could be finished and tested in short
time.
Perhaps, perhaps not. Maybe the patially implemented code that was
there turned out to not work and had to be rewritten.
>> What they did in last year and half ?
Mabye they finished it 6 months ago and have been working on IB 7 for
the past 6 months while QA was testing IB 6.5. <g>
David R.
J Hall
cstuntz@no_spam.vertexsoftware.com wrote...
Hi,
> > > XML support,
> >
> > Buzz-feature. Nothing more and nothing less.
>
> I agree. That said, there's nothing particularly wrong with this.
It worries me slightly, as IB has always been 'lean and mean'. I hope
the inclusion of such a 'throw away' feature in the engine is not a sign
of things to come.
J
J Hall
cstuntz@no_spam.vertexsoftware.com wrote...
Hi,
> > I was talking about the FIRST/TOP/LIMIT/Whatever syntax.
>
> ...Which is not an identical feature. Similar, perhaps, but there's no
> reason IB should constrain itself to MySQL syntax. The flavor IB uses
> is much more flexible.
Maybe, but I can only go on what is in the PDF's on the web site. I
shall wait till the SourceForge update before I comment further.
J
J Hall
cstuntz@no_spam.vertexsoftware.com wrote...
Hi,
> You don't understand how automated testing works?
> This release of InterBase has the
> equivilent of 15 years of testing performed, not to mention the beta
> cycle.
Automated testing is good, but it will only find old problems and/or
problems tests have been written for.
The best way, (and only real way IMHO), is for real users to use it in
real situations.
J
J Hall
Hi,
> > Which has meant they have wasted time and money fixing bugs that have
> > probably already been fixed, when they could have been spending it on
> > the ones that haven't.
>
> They can't just take FireBirds code and put it into IB . They have to approach
> the copyright owner of that code and ask them permission to put it in.
I am not a lawyer and all that, so I have miss understood the situation.
I suppose if we were just talking about Firebird and IB/OE there would
not be a problem, but as the commercial version is involved, that would
have 'issues'
> > This is one area where I think there should be some communication
> > between the two parties, there are enough outstanding bugs to go round.
> >
> I do not expect communications to ever start again.
There does not have to be any 'real' communication, just a gentlemans
agreement to 'bug fix swap'. For every bug 'taken' from the other
party, one is taken back.
I am sure two people from each side could manage to have a 'we will fix
that one if you do the other one' conversation without pistols at dawn.
Fixing bugs would hugely benefitt both parties, and once they are out of
the way, they can get on with the new features.
J
Bill Todd (TeamB)
years ago that on average, over several major releases of that product, beta
testers found less than 1% of the total bugs. I don't mean to imply that the
1% are not important or that beta testing is not valuable but rather to put
its contribution to finding bugs in perspective.
--
Bill
(TeamB cannot answer questions received via email)
Bill Todd (TeamB)
distribute it if you choose not to.
Jason Wharton
> the certified editions, because that (+ support) is what funds IB's
> development.
And that mentality is a precursor to stepping over a hundred dollar bill to
pick up a dollar.
What it boils down to is *when* you want to reap your harvest.
With InterBase I see massive potential that awaits a properly executed
strategy. If they cut off the market penetration and mindshare capturing
vehicles, InterBase will still earn (with a sure guarantee) that "dollar"
(virtually insignificant in comparison to Borland's other revenue generating
products).
But, if they capture and ride the wake of the new paradigm in software
development and dissemination, they will create a fortress of a product that
will easily topple and cut deep into the bloated pigs out there hogging the
database industry. If InterBase doesn't do this, mark my words, someone
else's will.
Would you rather make $1,000 from 1,000 customers for a million in revenue
and high expenses or make $100 from 1,000,000 customers for a 100 million in
revenue and less expenses?
The only thing the Firebird team is missing is a strong parent company with
a little "chump change" and infrastructure to keep it striding forward. The
only thing Borland is missing is a totally open, totally collaborative
development effort that involves and integrates a lot of people in the
development arena, as well as a way to penetrate in with no-cost barrier
product reasonably compatible with the value added revenue making product.
Just a thought.
Jason Wharton
CPS - Mesa AZ
http://www.ibobjects.com
Jason Wharton
> poaching.
I hope Borland poaches as much code from Firebird as possible.
Why should Firebird care?
Before you answer, think of the nice poach Borland afforded them....
Besides, that would be great flattery in my book.
Regards,
Jason Wharton
> problems tests have been written for.
>
> The best way, (and only real way IMHO), is for real users to use it in
> real situations.
That is the last place software should be "tested".
This is one aspect of Open Source development I totally detest.
My users and supervisors would skin me alive if I tried to adopt such a
mentality.
There will always be great value in properly tested software *before* it
hits the streets.
I know I will never use something where I work that doesn't have a properly
certified test process and guarantee of satisfaction and operability.
Don't get me wrong, there is a place for a community of developers even
outside traditional BETA circles which provide a massive testing base, but
nobody should ever promote the idea "end users" should be the proving
grounds of software quality. Things are too much that way already as it is.
Long live a certified InterBase and long-live Open Source InterBase. Let's
just not confuse where each is appropriate.
It's sure a good thing civil engineers don't use their users as the testing
grounds....
My $0.02 worth.
Jason Wharton
> year, while AFAIK Borland as any other commercial software company do
> testing at the end, so I can't be very far from truth that they tested it
> for cca last 3 months (at best).
I have to beg to differ here. No offense to the developers involved and to
the Firebird team in general but someone rushed FB RC1 out the door even
though it was broken in many usage scenarios.
Firebird's RC1 was was not tested thoroughly. I venture to claim it was
unusable. The cause of the problem was easy to retract but it is clear there
isn't yet a mature and certifiable testing process to guarantee against show
stopping issues out the gate.
I know this wasn't an official release but it was a disappointment
none-the-less. I expect better. So should everyone else.
Regards,
Jason Wharton
> Borland put the XML support in a separate DLL so you don't have to use it
or
> distribute it if you choose not to.
Thanks for that clarification. Highly useful information.
Max
"Jason Wharton" <jwha...@ibobjects.com> wrote
>
> I have to beg to differ here. No offense to the developers involved and to
> the Firebird team in general but someone rushed FB RC1 out the door even
> though it was broken in many usage scenarios.
There was a strong pressure from users (who don't think about compiling
their own engine from CVS) to release a binary version with certain new
features developed for 1.0 and not present in Beta2 build, because they want
to test them. RC1 is not a final product, it's beta. RC1 means only that's
"feature complete" and no new features would be added. Simple as that.
> Firebird's RC1 was was not tested thoroughly. I venture to claim it was
> unusable. The cause of the problem was easy to retract but it is clear
there
> isn't yet a mature and certifiable testing process to guarantee against
show
> stopping issues out the gate.
RC1 was tested internaly and in a small circle of users before release.
Wider testing (public release) shows some problems that were quickly
adressed. RC2 is around the corner. Would be RC2 considered a final ? No one
knows, but all users can participate on decission.
> I know this wasn't an official release but it was a disappointment
> none-the-less. I expect better. So should everyone else.
Jason, you're extend patterns of one methodology to different one. And it's
not true that RC1 isn't official release, just because everything that come
directly from Firebird project is an official Firebird release. It's simply
not a *final* or *stable* release, but anyway, any release in open source is
just a line in sand, and *stable* means only that *core developers* and
*majority of users* are satisfied with particular version. At the end,
*stable* in open source means usually higher quality level than in closed
source, due to higher expectation to *stable* release, and because
developers and users are those who define it, not a marketing department.
Regards
Max
"Craig Stuntz (TeamB)" <cstuntz@no_spam.vertexsoftware.com> wrote
>
>
> Your speculation is totally wrong. IB 6.5, as released yesterday,
> provides VLDB support for Linux kernels which support it (i.e., 2.4), as
> well as < 4GB DB support for kernels which don't, in a single version.
Back up your claims. I can't find anything about 64-bit I/O for Linux in
Borland's published materials. I can find only references to Windows and
Solaris, no word about Linux. Give me an URL of document where I can find
it!
> Actually, I'm impressed that there is no charge for the update beyond
> the media kit. I don't think Borland makes any money on the media kits,
> so this release is something of a gift to customers.
Actually, Media Kit is a nice source of printed documentation for
InterBase6/Firebird for reasonable price. I suppose it's a bestseller :)
Regards
Max
"Jason Wharton" <jwha...@ibobjects.com> wrote
>
> I hope Borland poaches as much code from Firebird as possible.
>
> Why should Firebird care?
If Borland would give back (somehow, even in advice), then why not ? But no
one likes "free riders".
Regards
Max
"Craig Stuntz (TeamB)" <cstuntz@no_spam.vertexsoftware.com> wrote
>
>
> You don't understand how automated testing works? Take a look at the
> InterBase Test Control Suites -- Borland released them to the public,
> but they can't release the test data used for IB (because it contains
> proprietary customer data). This release of InterBase has the
> equivilent of 15 years of testing performed,
I looked very, very closely :-) But did you ? If at least 5% of tests held
in-house has the same "quality" as average test published (around 400
tests), then I would not surrender any vital data to a product "certified"
by it <g> And that's just regression testing. Well, I would consider
"certified" InterBase as "good enough" for most operations, but it's just
that. Anyway, I would accept that level of quality when, and only when a
list of bugs fixed and still open would be available to general public (or
me :), and if I would like to be a really paranoid, I would demand sources
to look at them. <g> At last, database is not a RAD tool :)
> not to mention the beta cycle.
:-))) Do you know something that I do not ?
Regards
Robert Schieck
> Back up your claims. I can't find anything about 64-bit I/O for Linux in
> Borland's published materials. I can find only references to Windows and
> Solaris, no word about Linux. Give me an URL of document where I can find
> it!
>
Why do you care? You are wasting valuable resources (your time) that could
be applied to your favorite product FireBird. FireBird and opensource
projects are always looking for more talent to help with the project. I
suggest that instated of wasting your time here, you go over to FireBird
and help them work on the product to move it forwards.
Robert Schieck (TeamB)
MER Systems Inc.
Inprise/Borland/InterBase Search Engine http://www.mers.com/searchsite.html
Robert Schieck
> If Borland would give back (somehow, even in advice), then why not ? But no
> one likes "free riders".
>
> Regards
>
Borland open sourced the entire InterBase product, that firebird used to get
started with. They also put out the source for InterClient, IBX, and
IBConsole.
So when is the free riders called FireBird going to give something back
besides a bad attitude.
When are you going to tell us who you are besides "regards".
What really boggles my mind is why you are here. You could be spending your
time/resources helping FireBird with documentation and moving the product
forward instead of beating your head against a wall here. But then, that
would be constructive and productive.
the Poet
Craig Stuntz (TeamB)
J Hall wrote:
>
> In article <3C0E5F24.D00F48C7@no_spam.vertexsoftware.com>,
> cstuntz@no_spam.vertexsoftware.com wrote...
>
> >
> > ...Which is not an identical feature. Similar, perhaps, but there's no
> > reason IB should constrain itself to MySQL syntax. The flavor IB uses
> > is much more flexible.
>
> Maybe, but I can only go on what is in the PDF's on the web site. I
> shall wait till the SourceForge update before I comment further.
Note that the ROWS feature will not be a part of the Open Edition (for
this version, anyway), according to the press release.
Craig Stuntz (TeamB)
Jason Wharton wrote:
>
> It's sure a good thing civil engineers don't use their users as the testing
> grounds....
I agree. I'll go further and say that you can't test quality into a
product. Testing is primarily valuable as a check and balance to ensure
that other processes are working correctly.
Craig Stuntz (TeamB)
Max wrote:
>
> I looked very, very closely :-) But did you ? If at least 5% of tests held
> in-house has the same "quality" as average test published (around 400
> tests), then I would not surrender any vital data to a product "certified"
> by it <g>
The tests released were basically a demonstration of how to use the TCS
framework. They are *not* the certified edition tests.
> > not to mention the beta cycle.
>
> :-))) Do you know something that I do not ?
Not unless you were unaware that InterBase, like every piece of
commercial software I'm aware of, has a beta cycle.
Craig Stuntz (TeamB)
Max wrote:
>
> Back up your claims.
If my word isn't good enough for you, then you'll just have to order
the media kit and see for yourself. I'm amused that you would challenge
me on this.
David R. Robinson
>> users to use it in real situations.
> That is the last place software should be "tested".
> This is one aspect of Open Source development I totally detest.
Well said. I agree completely! This also contributes to why
corporations shy away from open-source software.
David R.
David R. Robinson
complete" and no new features would be added. Simple as that.
I disagree. In the industry, RC stands for Release Candidate. This
means that it is feature complete and a candidate for the official
release. If it's, as you say, just another beta release, it should not
be called an RC release.
Just my 2 cents.
David R.
J Hall
Hi,
> > The best way, (and only real way IMHO), is for real users to use it in
> > real situations.
>
> That is the last place software should be "tested".
I did not mean in production, but on development boxes etc.
> This is one aspect of Open Source development I totally detest.
Its just the same as closed source development, only much more public.
I have been on beta tests where the product has been really unstable and
buggy, but because it happens behind closed doors, it is never seen
> I know I will never use something where I work that doesn't have a properly
> certified test process and guarantee of satisfaction and operability.
I would never use anything in production that has not been fully tested
by ourselves, to our satisfaction. That fact that a product has been
through certified test process does not mean it is stable and/or bug
free (IB5.5, Delphi4 to name 2).
> Don't get me wrong, there is a place for a community of developers even
> outside traditional BETA circles which provide a massive testing base, but
> nobody should ever promote the idea "end users" should be the proving
> grounds of software quality.
At the end of the day it is down to the users. Whilst I am quite
prepared to use the latest 'bleeding edge' build of a product on may
development machines, and submit bugs etc. where appropriate; others
won't touch it till it has been 'released'.
> Long live a certified InterBase and long-live Open Source InterBase. Let's
> just not confuse where each is appropriate.
It is up to all of us to decide which versions of what software are
suitable for us. There are lots of factors to take into consideration,
from the basics of cost to the type of business the potential customer
is. I would not put a 'certified' version of a product into production
if the 'open source' version was more stable and had less bugs, unless
the 'end user' had any objections.
I would say that well over 50% of the software we use is un-certified,
'open source', this is partly down to budget, but in a lot of cases, it
is because the OS versions are as good as, if not better than the
'commercial' ones. An added benefit for all parties from using OS
software is there is more money left to purchase support contracts and
training.
J
J Hall
Hi,
> No offense to the developers involved and to
> the Firebird team in general but someone rushed FB RC1 out the door even
> though it was broken in many usage scenarios.
>
> Firebird's RC1 was was not tested thoroughly.
Well it was a release candidate, not a release. Besides, until it 'go
out the door' nobody apart from a handful of people could test it.
J
Jason Wharton
currently employed as a full time employee. I speak with that hat on and
largely from the perspective that my superiors, the decision makers, come
from.
I have nothing at all against the open source concepts as long as they stay
within the domain they belong in. I intend to convey ideas which may
eventually extend their domain to include where I work if they are ever
addressed, but for the time being it isn't so. I intend to tell you why
below.
In short, consider me the high profile corporate/government advocate. Don't
take offense, just consider my points.
> Jason, you're extend patterns of one methodology to different one.
Not sure what you are implying here but I don't think I'm off the mark in my
disappointment. To me, release candidate means something has passed all
internal and close inner circle testing constraints. If after this point
something hits the streets with show-stopper flaws in it then there is a
major flaw in the internal testing, period. I call a spade a spade when I
see one. (I have my screw-ups too.)
I am busy enough in my own programming realms and prefer to not have to be
the QA department for every product I make use of. I want tools I can make
use of to solve my own creative challenges, not have to solve the problems
in the tools I use in order to get on with my own issues.
Thus, it is of great value to have something considered much more than a
line in the sand which becomes something I can take great confidence in,
without having to know the entire intricacies of the database's internal
implementation and who has been monkeying around with it over the past few
months. No matter how much I may appreciate what is done.
> And it's
> not true that RC1 isn't official release, just because everything that
come
> directly from Firebird project is an official Firebird release. It's
simply
> not a *final* or *stable* release, but anyway, any release in open source
is
> just a line in sand, and *stable* means only that *core developers* and
> *majority of users* are satisfied with particular version. At the end,
> *stable* in open source means usually higher quality level than in closed
> source, due to higher expectation to *stable* release, and because
> developers and users are those who define it, not a marketing department.
I can appreciate the open collaborative development effort has excellent
benefits in many areas. I am not in the slightest opposed to this aspect of
things. But, I find it appalling the attitude about this "line in the sand"
mentality. The burden it imposes on people in my position to have properly
certified tools to work with is totally unacceptable.
In short, it is simply because there isn't enough motivation to bear down
and grease the testing wheel and put a sufficiently long halt to the
development wheel for something to reach a point where it can actually be
considered a certifiable release. In my own experience this is a difficult
thing to disciple myself on.
In general, because OS releases are so frequent, it fragments people's
understandings so much that the only way to take any significant measure of
confidence in something is to be a tightly integrated part of the actual
development process. That isn't attractive, affordable and practical for
many, if not most, development environments. It makes support and product
integration very cumbersome as well.
This "line in the sand" approach is for people who are in academic pursuits
and don't have real jobs with real responsibilities. The things which I am
responsible for have to do with presidential elections for the United
States, filings over which any mess-up could lead to multi-million dollar
law suites, and so on. I know you want to blast me with all the scenarios of
companies who are taking the plunge. I consider them either able to afford
to do it right or foolish. I don't want to be foolish and our agency is not
in a position to do it right.
The "line in the sand" mentality just doesn't belong in my work environment.
When the OS movement completes the cycle of proper software development into
a nice round wheel, then I may venture to take a spin with it in my work
place. But, from what I can tell, unless I want to be their QA department, I
hadn't better take the ride because it is just too bumpy for me.
With all due respect.
PS. I do have Firebird at home and enjoy using it there and hope to
contribute what I can, I just know it won't happen where I work without a
major attitude change in my superiors. And I am not going to disagree with
them because I clearly see their point.
Regards,
Jason Wharton
Government Employee
J Hall
Hi,
> To me, release candidate means something has passed all
> internal and close inner circle testing constraints.
This is where I differ. To me, release candidate means 'feature
freeze', and is a point where no new features are added and the serious
work of testing and bug fixing begins. We use this methodology in-
house.
J
Jason Wharton
> > internal and close inner circle testing constraints.
>
> This is where I differ. To me, release candidate means 'feature
> freeze', and is a point where no new features are added and the serious
> work of testing and bug fixing begins. We use this methodology in-
> house.
Ok, we are just hung up on semantics. The reason is because of the
difference in mindset we have.
In the commercial world of professional software development, a release
candidate means the bugs are all rooted out to a certifiable level and for
all intents and purposes, it should be ready for live,
production/mission-critical usage.
With gross exaggeration and a little humor to highlight the point, in your
mindset it means the monkeys have finished playing and it is time for the
community to clean up the mess.
How far off am I?
Where you will make a good defense is that humans are involved in all
processes here and an open air collaborative environment tends to bring
things to the surface faster than a select few under one roof.
My only point is, I want my cake and to eat it too.
In other words, I'm not sure I have one.
Just more grub for thought.
Jason Wharton
> > > real situations.
> >
> > That is the last place software should be "tested".
>
> I did not mean in production, but on development boxes etc.
>
> > This is one aspect of Open Source development I totally detest.
>
> Its just the same as closed source development, only much more public.
> I have been on beta tests where the product has been really unstable and
> buggy, but because it happens behind closed doors, it is never seen
Now we are getting onto the same page, sort of.
> > I know I will never use something where I work that doesn't have a
properly
> > certified test process and guarantee of satisfaction and operability.
>
> I would never use anything in production that has not been fully tested
> by ourselves, to our satisfaction. That fact that a product has been
> through certified test process does not mean it is stable and/or bug
> free (IB5.5, Delphi4 to name 2).
I agree one should do their own testing. It is very possible I could use
someone's product in an inappropriate manner without knowing it. It's the
idea of a "vendor" expecting me to do *their testing* that I am not
comfortable with. I am much more comfortable in an environment where you
operate from as few variables as possible. The more variables in an equation
the bigger the mess on your hands.
The ideal is to have something that works exactly as advertised and isn't a
moving target.
> > Don't get me wrong, there is a place for a community of developers even
> > outside traditional BETA circles which provide a massive testing base,
but
> > nobody should ever promote the idea "end users" should be the proving
> > grounds of software quality.
>
> At the end of the day it is down to the users. Whilst I am quite
> prepared to use the latest 'bleeding edge' build of a product on my
> development machines, and submit bugs etc. where appropriate; others
> won't touch it till it has been 'released'.
I don't see how your first sentence above relates to the rest of the
paragraph.
> > Long live a certified InterBase and long-live Open Source InterBase.
Let's
> > just not confuse where each is appropriate.
>
> It is up to all of us to decide which versions of what software are
> suitable for us. There are lots of factors to take into consideration,
> from the basics of cost to the type of business the potential customer
> is. I would not put a 'certified' version of a product into production
> if the 'open source' version was more stable and had less bugs, unless
> the 'end user' had any objections.
>
> I would say that well over 50% of the software we use is un-certified,
> 'open source', this is partly down to budget, but in a lot of cases, it
> is because the OS versions are as good as, if not better than the
> 'commercial' ones. An added benefit for all parties from using OS
> software is there is more money left to purchase support contracts and
> training.
What it boils down to is trading money for time. Those who can, do. If a
company has money to spend they are going to much favor plopping down some
cash to get high quality tools so that their business plan can be quickly
implemented and in operation with accuracy and robustness. If a company
doesn't have lots of money and they have access to cheap labor, open source
models are the way to go. Each have their trade-offs.
For one, I prefer the model which doesn't shape the mindset of the software
industry into thinking software developers efforts should be a free-gratis
affair. I believe the talents I possess to be of great value and I like to
see a healthy mindset in the market for them. Don't get me wrong, I believe
in good-will, but it baffles me to see so many willing to prostrate
themselves to the "free-beer" software movement when is all they are doing
is sabotaging the foundation of their income potential in the "big picture"
of things. It's like programmers are in a mad dash to work themselves out of
a job and not even make a living in the process. <g> Where is the sense in
this?
Grub for thought.
Regards,
J Hall
Hi,
> What it boils down to is trading money for time. Those who can, do. If a
> company has money to spend they are going to much favor plopping down some
> cash to get high quality tools so that their business plan can be quickly
> implemented and in operation with accuracy and robustness. If a company
> doesn't have lots of money and they have access to cheap labor, open source
> models are the way to go. Each have their trade-offs.
It can also depend on what type of software you are developing. Most of
our development if for other people, and by having the option of using
open source or a 'certificated' version of an open source product gives
us a bigger target audience. Some people have the money, some people
don't, some companies and government departments would entertain using
anything but certificated, paid for, software, others would not by our
product if they had to pay a per user/server licence fee. Its all down
to options.
J
Jason Wharton
> out the door' nobody apart from a handful of people could test it.
My point is (and I actually hope to make some nice ones as my own
contribution to the cause) there needs to be a comprehensive level of
regression tests to uphold a certain baseline of insured quality. I believe
this is a critical element and it is one thing that Borland has and does
that I like.
Before something goes out the door is it a very valuable thing to know that
you have assured certain things "still work". That is worth paying some
money for in my opinion.
I do also agree test suites can hamper progress too. What if the test suite
has a bug in it? That can get ugly too.
I remember working on the 2nd largest POS (point of sale) application in the
USA for a few years. 12 developers and 5 QA on our core team. It was like a
war zone between dev and qa. In the end run of things, assuring quality is
critical and the more automated with accuracy it could become the better.
It's a mixed lot of pros and cons...
The one thing in favor of Open Source is you notice the only really
successful projects have users who are themselves programmers.
Claudio Valderrama C.
news:3c10211f$1_1@dnews...
>
> In the commercial world of professional software development, a release
> candidate means the bugs are all rooted out to a certifiable level and for
> all intents and purposes, it should be ready for live,
> production/mission-critical usage.
Please name five commercial products that you unpacked, installed and were
free of any bug, installer included.
Second, you probably know that most commercial products say their warranty
fades after 30 or 60 days and that they CANNOT be used in mission-critical
scenarios, specially involving risky activities, medical requirements, etc.
You read everywhere NO WARRANTIES.
Third, a company has good and bad developers, even with a closed commercial
program. An open source effort has mostly employees with a regular work on
their respective jobs that donate time, so you get almost the same spectrum
of people than inside one single company. I don't know any proven fact that
one group is superior to the other. It coul be said that contributors don't
care what they do, because they don't get direct income. However, at least
in the case of FB, most contributors are users. If I make a big mistake, I
will eat the pie in my own application when using FB. On the other side,
with the same bad faith, one could say that regular paid devs do only what
they are paid for.
Also, not much commercial applications have a detailed log of current and
fixed bugs that the customer can read and understand. If I learn that bug
450-89 was closed and bug 453-38 isn't reproducible, I'm left where I
started.
Pretending that a commercial program is crash-proof just because it was
certified is as utopian as pretending that an open source project is perfect
because there are many eyes looking at it. A commercial program is focused
on features. This makes the day of the marketing department. An open source
project is focused first on bug fixes and cleaning code unless all
developers throw features and forget the other side. For a mixed example,
IBO is trustware, so it should be in the middle of commercial SW and open
source SW. Should I imply that it's bug free? I can't, because we are all
humans.
The release date for a commercial program is set in stone unless a very
serious bug is discovered (even so, several marketing depts will keep the
release date fixed at any cost and one day after the release date you are
already installing patches that are 50% the size of the full application).
The release date for an open source program should balance contributors'
desire to clean this or that, mend this glitch, add "this small feature that
I have in my disk since 4 months ago", keep the "infinite refinement cycle"
running, reproducing the issue that an alone user found in the South Pole
at -45 Celsius, please the independent devs that want a new freeze point to
try and satisfy the pressure of some people and companies that want a stable
release to ship their applications with it for users that don't care or
can't afford an equivalent commercial product.
C.
--
Claudio Valderrama C. - http://www.cvalde.com - http://www.firebirdSql.org
Independent developer
Owner of the Interbase® WebRing
Martijn Tonies
a
> moving target.
That'll be the day.
:)
--
Martijn Tonies
InterBase Workbench - the developer tool for InterBase and Firebird
http://www.interbaseworkbench.com
Upscene Productions
http://www.upscene.com
"This is an object-oriented system.
If we change anything, the users object."
David R. Robinson
were free of any bug, installer included.
There is no such thing as bug free software (commercial, closed,
open-source, free, whatever). Period.
>> Pretending that a commercial program is crash-proof just because it
was certified
I don't think that's what Jason is saying. The whole point of this
sub-thread was that Firebird released a "RC" (release candidate) build
that had some serious problems in it. All Jason was trying to say is
that in the commercial world, RC means that the build has had extensive
and thorough testing and is a candidate for the official release.
That's obviously not what RC means to others.
>> The release date for a commercial program is set in stone unless a
very serious bug is discovered
Some companies operate this way some don't. The entire software
industry known for never hitting the release dates.
David R.
Ian Newby
I suppose I should give some information as to why these things
interest me (and then you can tell me a better way to do it <g>).
We currently store a number of scripts in our version control system
as it allows to to check changes between versions.
There are three main scripts:
1. Database creation.
2. Basic lookup data insertion.
3. Test data insertion.
These three scripts are booked out of CVS and a development/test
database is built (currently by hand).
The product is at a very fluid stage in its development so database (and
test data) changes occur very frequently.
So consistant metadata extraction, script running (and the ability to
batch load various amounts of data) are very dear to my heart.
Regards
Ian Newby
Jason Wharton
This whole thing is a paradox. Your truth is invalid without my truth and my
truth is invalid without your truth. The complete truth encompasses all of
it and then some.
The point was about the semantics and domains. If you don't see my point and
the value in it, that's ok. I see your point and see the value in it. It
just isn't something the culture I work full-time in accepts, and for
justifiable reasons.
This is all completely beside the point of my personal interests and my
interests in supporting Firebird with the product I have made. Firebird is
great in the domain in which it is best suited to serve.
In short, there is a rightful domain for a commercial certified InterBase
and there is a rightful domain for Firebird. There isn't a whole lot of
overlap and the only part I would like these two domains to recognize is
that each composes an integral part of a greater whole.
In short, and this is my vision of things, if the two groups insist upon
separate self-consciousness they will each roll off with a huge piece
missing out of their "wheel of progress". Anyone who has studied the
dynamics of physics and resonance will know that any imperfection in the
rotating object will result in foul harmonics of the system and it will
eventually destruct and be broken down into chaos if does not stay within
its inertial constraints. I'd like to see those constraints removed to a
much higher degree. It may happen yet....
FWIW,
Pete Melissakis
Anthony Hopkin
my
> truth is invalid without your truth. The complete truth encompasses all of
> it and then some.
>
It sounds to me as you have been listening to my Tai Chi master. [;¬)}
Keep up the great work Jason!
Wildly great products - and a philosopher to boot. No BS!
happy days
Anthony