Open-Sourcing Interbase - Ed Popkov
What the IB Community Thinks About Open-Sourcing Interbase
David Intersimone "david i" Vice President, Developer Relations at Inprise/Borland asked Robert Schuff this question:
What if we open sourced Interbase 6 for all platforms?
what do you think about this idea?
I think the Linux community, at least, would love it.
Your thoughts would help add to the discussions going on internally.
Robert sent his answer back to David i and copied it to the community on the interbase.mers.com list. Here is how the community responded.
Robert Schuff: Well, ultimately the only thing that matters to me is that development continue with InterBase and that InterBase continues to be available (ie I can purchase additional licenses. As far as support goes, I have never seen the need to purchase support from Borland/Inprise for the product.
So with that said, I can say that open sourcing IB6 potentially could be an excellent thing, or it could be a disaster for InterBase. It appears to me that the wild card with open sourcing interBase (as compared to the Linux model) is that InterBase is cross-platform. So what I would worry the most about, is that compatibility across platforms could be lost. This is a major advantage of interBase and I would hate to see it go. I do not know much about open sourcing and management of the code, but I wonder if it would be possible for there to be some kind of IB steering committee whose charge would be to set some standards for IB development and maybe there would even need to be a certification scheme so that I know the interbase I am installing would behave in a ceertain way.
Within the constraints above it seems open source would be awesome and would let interbase develop into the RDBMS it **should** be.
In my mind the worst thing that could happen would be for Borland/Inprise to just stop development of IB, continue to milk money out of v5.6 and make it impossible for anyone else to take the product and run with it. I think that kind of situation would also be bad for Borland/Inprise in terms of developer relations and most certainly would be seen as morally reprehensible by the IB community. There are a number of people who have literally staked there lives on interBase (mortgaged houses, maxed credit cards, etc) in order to claim their share(based on IB). To leave them (the Frank Borland's if you will) with no way to viably support and develop their products would be against what we as developers have thought Borland was all about.
I urge you to continue to be a champion for the developer (as you always have) and make it known to those who would not otherwise what the impact of Borland's decisions will be to the loyal (but very frustrated) Borland developer community.
Reed Mideke: Most cross platform tools are built from a single code base (with a few platform specific files). Futhermore, most of the successful open source projects are VERY cross platform. Linux itself runs on a large number of achitechtures, and things like perl and GCC run on even more. The point here is that for a commercial product to be ported to a new platform means that there has to be an immediate promise of revenue.
For an open source project to be cross platform only requires a motivated developer with access to the desired platform.
If the license used is like the GPL or LGPL, then anyone who distributes the the product MUST make their source available. This ends up being a strong detterent to 'forking', since any changes that are seen as generally desirable can be ported to any version. Historically open source projects tend to be the most interoperable, portable, solutions around. I personally think that releasing InterBase under the LGPL would be an excellent idea. If you have input on the type of license used, you should shy away from BSD/Xfree style licenses, and push for one which
Since InterBase users make their living selling solutions based on InterBase, not being able to make their own proprietary versions should not be a disadvantage. Having all changes available to everyone will help keep the playing field level.
- allows you to use the source freely
- requires you to publish modified source
- allows linking with proprietary code
As for management, I would hope that an 'interbase.org' would arise, either supported by Inprise, or by those who have a large investment in InterBase. This group would accept contributions, and release a 'reference' versions of InterBase. This would be much like the 'linus' tree of Linux. Many linux distributions do not use a kernel straight from the linus tree. They apply their own chosen patches, but since they always start from the linus tree and apply their patches, no significant divergence happens. If the interbase.org could hire (or otherwise get input from) some of the senior interbase developers to provide input/review on the submissions, I think it would have a very good chance of success.
IMHO, the key to a successful open source project is a large body of motivated developers, which the members of this list clearly represent.
> Within the constraints above it seems open source would be awesome and would
> let interbase develop into the RDBMS it **should** be.
I agree. I think mature, open source RDBMS has HUGE potential, indeed, far beyond what commercial InterBase ever was. This would also create greater opportunities for those who are already selling solutions based on InterBase. If the license is relatively open (compatible with the Open Source Intiative's Open Source Definition) I think that there would be a lot of support from the open source and linux comunities as well. I'm sure that the Linux businesses (Red Hat, SuSE, VA Linux etc.) would see the benefit and provide some support.
I encourage everyone who has in investment in InterBase to make their wishes known (politely, of course). Responding to the "COMMITED TO INTERBASE" thread or Dalton Calford's survey is a good way of doing it. Writing your specific interests to David I is also a good idea.
Joseph Alba: An Open Source Interbase will be worth much more in terms of marketing dollars, than a Closed Interbase 6 (or a non-released Interbase 6). A basic principle in Market positioning is "Be Different". Open Sourcing an excellent product, backed by a major Software Company (Inprise) would be so shocking that it would create a stir, and direct attention towards Inprise and its products.
(If there is some reservation towards open sourcing, maybe a compromise could be, just Open Source it on Linux, and the initiatives done there can be easily ported to other software platforms by Inprise developers.)
Another suggestion, which ought to have been quite obvious is: Release the SQL Link of Interbase for BDE as a freely available product / download for all versions of Delphi, etc...
I mean, it is quite obvious for a company to give its own product an "unfair edge" over the rest, right? So, if SQL Links for other products are for enterprise editions only, Interbase SQL Links should be freely available so that those having Professional versions won't have to languish with ODBC's and those who do benchmarking won't have a valid excuse to slow down Interbase by using ODBC.
>Within the constraints above it seems open source would be awesome and would
>let interbase develop into the RDBMS it **should** be.
This is the crux of the matter, isn't it? I mean, the real question is: what _should_ an RDBMS be?
To me, the answer is that an RDBMS should be fully compliant with SQL/92 (or SQL/99). That is, it should be standard both in its treatment of SQL and in the call-level interface (CLI). As yet, there is no RDMBS that meets the standard. There should be at least one.
...I am skeptical of any open source approach to RDBMS. What, essentially, is in it for end user developers? Certainly, open source will cause change, but what kind? If SQL and the API are extended and/or modified again and again, then open source will simply act to proliferate the variety of development interfaces and so make our jobs ever more difficult. If, on the other hand, there was a strict discipline on standards, the change may allow us to extend our applications everywhere.
I don't know what it is that kept IB so close to the standard for so long, but I am grateful for it. Human nature begin what it is, as soon as IB went to open source I would back off. It would be a long time watching and listening before I'd risk my business on it and a long time after that that I would become enthusiastic. God save us from everybody's good ideas!
Rea Berryman: I am NOT in favor of open sourcing IB.
I am in favor of an independently run IB that is organized and run as a web based company. I believe that it is very important for the existing IB development team to be kept intact. This will allow those most experienced with the development of IB to continue with their work and to move forward with the release of IB6 and with the planning of IB7. I really do not want to see Inprise redirect the IB team to other endeavors.
I would like to see Bill Karwin and company return. I have been very satified with the steady progress and vision of the current IB team and they have my support to continue doing a great job.
I would also like to see the IB source code made available to developers who are willing to sign a nondisclosure agreement. This would allow a lot more people the opportunity to fix bugs and speed up the addition of features.
It is my belief that a lean and mean web based IB company would be very well supported by the IB community. Let's face it, real developers are passionate about their work and live to improve upon what they did the day before. This is evident in all of the fine work the existing IB team has been preforming.
If Inprise will consider helping IB become an independent company, then I would like to see if the IB community could rally to help make it a reality. Open source can always be done later if things don't work out.
Stephan Patterson: I don't believe that open sourcing Interbase would work well at all.
My preference is to see Interbase become an independent software company. As users of the product, we want the company to be free to pursue all opportunities available to it, whether or not they are compatible with Borland's vision of the day.
I wish Bill Karwin and the others who resigned last week could return, although I doubt this is feasible at this point.
I am sorry that we need to have this kind of discussion again. Users of Oracle or SQL Server are not debating whether their database of choice has a future. Even if things work out, there's always going to be a doubt as to what happened that Friday that prompted so many people who obviously cared for the product to resign. In the information age, I regard the choice of a corporate database as even more critical than that of a programming language or operating system. Given its negligible market share, the setback caused by the departures and the uncertainty regarding its future as a Borland database product (especially with JDataStore in the picture), many will find it very hard to justify a commitment to Interbase.
Claudio Valderrama: I don't have direct experience with Open Source projects, but IMHO I would prefer an independent company or organization to lead the IB developments.
Problems: those who even stay with Borland and those who resigned (like Reed Mideke, if I'm not wrong) have experience making IB compile/run in the N platforms the product supports. This is not the only expertise that must be almost learned completely again if IB goes free, wild Open Source and Borland doesn't compromise any more effort.
At least, if we are compelled to go full source code, then there must exists a reference committe or Interbase.org, hopefully with the cooperation of the former ISC employees, to make the project successful.
Please don't get me wrong. I'm not putting in doubt the IC and skills of many independent developers who love source code à la GNU. My concern is not to lose the promise of great compatibility among the different platforms where IB is currently ran. The other concern is not to miss the boat with several features that run in only one platform. Because IB seems not to be written in ANSI-C, you can't expect the code for let say Linux to be copy-pasted to Wintel and work. As Jason Wharton said, IB has a legacy of several years in the market and there're things (features) here due to the age of the product. Again IMHO, perhaps the code may be "cleaned" a bit. Please, ISC employees, accept my apologies if this sounds offensive. I didn't want to say you maintain ill-formed code. I only wanted to mean there may be still code stemming from the birth of the product and some features or API calls that can be "remodeled" or streamlined.
There's an NDA hanging over my head. If Borland stops IB and donates or sells it to the public community, then us the IB6 beta testers will be able to speak and tell you (those who didn't get involved in the testing) why it's very important and beneficial that IB6 hit the streets.
Olivier Mascia: Though open-sourcing IB may seem a nice idea, I think we should all think twice about it. Yes the Linux community would be delighted about it. And for sure, IB could have a future there. The risk is to "linuxize" IB and loose cross-portability on other platforms.
As a developer and a company dedicated to long-term relationships with my customers, I need a product that fits my needs as perfectly as possible (IB is that), but a product that does not wear a "I'm dead" label or may wear such a label in some short time-frame.
Today I and my company *NEED* IB Version 6 to be released, at least on Windows platform, and later during 2000 on Linux platform. These are our own personal needs. IB 5.6 is an excellent product, but is missing some key functionnalities that IB 6.0 does offer, as it is today and as it could be released today (can't say more - NDA).
I can understand that keeping IB is becoming a little embarrassing for Inprise. Not exactly the kind of product on which the company seems to be willing to focus, today. These are suppositions on my side. One have just to look at the Inprise web site to see all that Linux/Java fever.
Open-sourcing IB may look like an "easy" way to get rid of IB, while may be not hurting too much the IB community of users and developers. My position is to say OK, IF AND ONLY IF, Inprise puts up important resources to help coordinate that open-source effort. But maybe the resources and attention needed to play that "federation" role in open-sourcing IB would still be considered a burden on Inprise focus to other technologies...
So let's see which other options are there. If the development cost of IB on such many platforms as it runs today is the culprit (I'm just guessing), why not evaluate the possibility of keeping IB inside Inprise, push its development further, but reduce the number of platforms on which it is available ? (oops I know I'll get flamed here, but I need to express that idea). Why not stop further development of IB on some platforms (just support current customers on those platforms) and push the development further on a limited set of OSes ? To fix ideas (and knowing I'll be killed by many people here) : why not focus ONLY on Windows and Linux versions of IB for the future of IB ? I do not want to diminish the importance or value of other platforms for which IB is ported today. I just want to find a proposal that allows Inprise to lower its development and QA costs in continuing development of IB. Continuing on those two platforms (Windows and Linux) may be the least hurting solution. A huge number of users/developers would be fine with this dual-platform choice. Windows will not disappear tomorrow. It's still good to bet on it. And Linux is becoming more and more important. (Maybe add also Solaris to the list if you have strong cross-relations with Sun, which I believe).
To synthesize, my proposal would be this :
Let the developers continue to be delighted by the goodies of Interbase, let the developers continue to produce excellent end-user software embedding your excellent database engine and those developers will sell that engine for you (just a matter of putting up some more dynamic in the VAR relation-ship with them).
- IB is of high technical value. Do not take the risk of diluting this asset in an uncontrolled open-sourcing effort.
- Value the IB asset by dropping new development on a number of platforms, and clearly committing to pushing further development on some reduced set of platforms, namely why not Win32/64 and Linux. That means some customer looses and some hurting. Most will be happy though.
- Promote IB as very valid DBMS solutions, highly supported by Inprise, and available on all platforms for which you have development tools (so Windows and Linux). This DBMS is not the "one size fits all" panacea. But it is a really good piece of software for the numerous programmers and companies around the world that wether do not require or cannot afford the operational complexities of major DBMS players like Oracle, SQL Server adn DB2. Very high end solutions will go to one of those players, for sure. But let others use something affordable, and very competitive on smaller project scales. In each country on earth, there are far more "small accounts" companies than larger ones. So far more prospective customers for IB than for other players. It's all in the way of selling the thing.
- If further effort is to be done to lighten the development of IB, drop add-on development and focus on the engine and its interfaces. There are already numerous excellent third-parties tools on the market for IB. Put them all on the IB CDs as evaluation versions, let the developers make their choice of those third-parties add-ons, and concentrate on the important : the excellent engine and its APIs and interfaces.
- No, I do not even imagine in my fooliest dreams that JDataStore could be a replacement of what IB is today. Maybe in 5 years or more. If by that time Java is still what it tries to become today.
Long live Interbase !
Michael Lant: I'm afraid that I can't support the open source initiative. In spite of the technological superiority of InterBase, we often encounter a great deal of resistance with our customers when we are selling our vertical market applications. The typical refrain goes something like this: "We are an XXX (insert any RDBMS) shop and we won't consider a product that will not run on that RDBMS." Most of the time we are able to overcome this obstacle because InterBase has a long history and is a well supported product. Although our product would definitely be superior product running on InterBase, the ideological and technology issues are greatly diminished by the issues of sales, support and marketing. The bottom line is that our customers (some 2000+ seats project for 2000) will simply not buy our product if it is based on an open source database.
Rea Berryman wrote:
>I am NOT in favor of open sourcing IB.
>I am in favor of an independently run IB that is organized and run as a web
I agree with Rea. Using something like Sun's community license is the way to go!
Claudio Valderrama: I have to agree with Olivier [Mascia] in some points. My clients mainly use Windows. Why? Because they're small and use PCs and not Sun workstations. Of course, I have some major clients, but they are a few and even some of them use Windows. Ok, small companies can use Linux as well but remember you cannot replace one platform in one day without causing headaches if the employees are not very apt to computers and at least are accustomed to Wintel.
I rely on some IB6 features to be able to release my little niche products in the next year. If Inprise gets rid of it now, *probably* the release would be delayed some months more because people need to organize and start understanding the code if IB goes free source.
My personal opinion is Inprise should release at least the IB6 versions that come in the beta CD-ROM, to know: Windows and Solaris. They're almost finished. Rough edges are minor IMHO. After that release that may be perfectly in February, Inprise can go ahead and release the full source code, reorganize the way IB is sold, span an independent company, etc.
I recognize the fact a product doesn't exist until it's released, but this is a formal, official statement made I guess because companies need to be sure they won't be sued if they decide to stop a beta testing program and kill the announced product. In this case, the situation is not as easy: Inprise cannot risk losing the confidence of its customers and IB is a product with 15 years (or more, correct me) of successful history. When I use "successful", I don't mean top rates sales nor great incomes nor 20% of market share but that the product was able to let developers do what they needed with their applications and in a very elegant way. Now, in the dark point, things could be better in the financial side if Inprise had had any muscle and intelligence in marketing the product.
I was surprised some months ago when I read a statement from Oracle's CEO (Larry Ellison if I'm not wrong?) where he envisions a great opportunity in the embedded market for specific applications so he wanted to position the smaller Oracle version in that "growing" market. So, are Inprise and Oracle watching at different markets or it's simply a fact of pessimism v/s optimism??? Will Oracle and $ql$erver be the only available option in ten years more?
My clients don't have HW resources nor money for Oracle. Only one of them uses it. The others use IB or MS-sqlServer and in the case of the micro-companies that are growing, they found that simple Excel spreadsheets cannot solve all their needs. So I offer little specific products able to work stand-alone or networked and really the server doesn't need to be in Windows like the client side (Delphi). IB saves my day in these cases for an affordable price. And at least in my country I have special prices for developers so the my clients can get a discount if they acquire my product along with IB instead of purchasing IB licenses for theirselves.
I think ISC needs to be converted in a separated company, be it traditional or free source based. And I expect a good help from Inprise in the first months if this is what will happen. Perhaps, an arrangement can be made: Inprise gives some funds and help at the beginning to the new company and get in exchange the right to distribute the product with their SW. The company can live from licenses and from technical support... although I suspect IB never has needed too many tech support.
On the other side, I remember some Netscape developer moaned some months ago that the idea of putting Netscape on the source code train was not very beneficial, because the vast majority of the work was still undertaken by the AOL employees (former Netscape employees). Although there were several independent developers, the real coding effort was still on the original employees. Also, I don't know how my clients would react if they know IB is no longer backed up by Borland but it's maintained by a group of happy friends from the Internet.
I have a good experience to share. In my country, much of the real-time work in the custom-house is done by hand and converted lately to computer information. I created a small app to help streamline the process. Because several employees hardly use Word or Excel and in some installations, the DB server would be shared and in other places, it would be ran in the same PC as the user, I didn't want to bother with those custom options, so I gave them one set of diskettes with my app and another set of diskettes with the IB4.2 installation. This is because some places have really old computers w/o CD-ROM readers while other places use multimedia Pentiums.
All of the 30 computer-unexperienced employees were able to install IB without further help. Embed, deploy, relax and I didn't embed really. I used the default setup. Can you afford the same with any other RDBM? And because I used IBO, I only had to install the db, detect where it's located and my program was working. Now, the inspections and documentation information is entered only once and keep in sync at evenings, but w/o tons of paper. And even the most humble PC was able to run my D3.02/IBO app with IB4.2 sitting on the tray as an icon. If they complete the future project of wiring all points of work and have Linux or Solaris based centralized server, the cheap app still will be able to work remotely provided they put IB4.X or newer on their UNIX-like server. The only point that required some effort and explanation was to enter the license codes for IB.
IMHO the critical platforms are Linux, Solaris and Windows. Ok, someone will claim that I must include Novell, too. Don't want to comment on relative merits of these platforms, I'm really saturated of the discussion whether Windows or Linux is the best OS ever. I don't care. I simply use the best suited OS for my needs, depending on the available resources.
Ramon van Alteren: I'm NOT in favor of open-sourcing Interbase... I think most of the arguments posted here before against open-sourcing Interbase apply for me as well. Cross-platform availibility and development would be a major issue when open-sourcing Interbase. Apart from that I think that the current client i'm working for won't agree to it and move to a different database........ :-(
Reed Mideke: As I've stated before, I think that the above are non-issues. Both would most likely be IMPROVED by open sourcing IB.
[that the current client i'm working for won't agree to it and move to a different database ] That is a valid concern. If a large number of customers would be unable to sell there solutions because IB was open source, then clearly, open source is not the best thing for the existing customers. I think that what you're really saying though, is that they would move to a different DB if no legal entity supported (meaning both develop and provide tech support) IB. This is a different issue, and is independent of the source being available. Would you're customers still move away if open source IB was managed by Inprise, or by a company which based its income off of providing support (ala LinuxCare etc.), or by a non-profit organization like apache ?
Veli-Matti Hurskainen (VMH): If open sourcing were a solution to anything, why it is not being done with every software package? IBOjects, Marathon?
Reed Mideke: Because it's hard (though not impossible) for the owner to continue making money off of it if you give the source away for free. But if the owner has already decided that they are not making enough money from it, then it may be more sensible/beneficial to the customers than just stopping development.
It is not clear whether or not Inprise thinks they can still make money off of IB, although the fact that people like David I. are publicly mentioning open sourcing it might be an indication. Selling the code to some one else is also a possibility, but finding investors for a brand new company that proposes to sell a virtually unknown SQL RDBMS is a shaky proposition, since it doesn't reasonably have the potential for the 1000% return of some crazy internet startup.
Joseph Alba: Again, there seems to be a misconception that Open Source means Freeware. Wrong. IB 4.0 for Linux is Freeware but not Open Source. Delphi's VCL and other libraries are Open Source but NOT freeware.
Open Source simply means that the code can be revealed along with the executable. So that if things happen to be broken, the user himself is invited to take a look and even suggest changes to the Owner of the code.
Unix has been Open Source and it made a lot of companies rich. Xerox open sourced its networking code, and this eventually led to Novell.
Throughout history, Open Source has led to the birth of something even bigger. That is why I think Open Source is a mighty idea for Interbase. Again, Open Source is not Freeware.
VMH: Linux was created because - and only because - it was open source from the very beginning. So is other GNU stuff. And it is ok. But they start on a different basis compared to Interbase. Interbase was there at the time Linus Torvalds went to primary school.
Reed Mideke: I think the biggest difference here is that linux is successful (in terms of having active development) to large extent because of the kind of person Linus is. Independent of his programing ability, he has shown himself to be a very good arbitrator of contributions. Whether there are people able and willing to fill such a position for IB is an open question, and would depend on what sort of organization developed around it.
VMH: The developer community around Interbase is very valuable to ISC (or whatever it will be) giving the directions where to go with the product, beta testing (hopefully it is beta, not product like it was with 5.0 and 5.11), creating third party solutions like IBO, supporting each other via this list (and it's succeeder), marketing the product and - most of all - using it in their own software products. This community is not - necessarily - headed for dbms business, it is a bunch of solution providers and as such more users of Interbase than it's developers (I'm sure there are exceptions to this). If mr. Menuhin had to build his violin, he'd never played.
Reed Mideke: Most of the IB community probably does not have the time to invest in continuing the development of IB. But, if even 1% was Interested enough to get there hands dirty with serious development, IB would have a much larger team than it ever had before. If another 5% was competent enough to find and fix some bugs when they ran into them, quality could be greatly improved. And it's already a pretty reliable product. Being that the DBMS is the foundation of your business, being able to directly improve it, and fix bugs that appeared with respect to your applications might be a big benefit.
VMH: If (IF) ISC - or whatever - needs new investors, developers, VARs, I'm sure this community will be a good resource for those.
Reed Mideke: I'm not say that open source is the best or only solution. Just that it is certainly preferable to completely abandoning the product, and has benefits and disadvantages compared to a new closed source business based on IB.
As I mentioned in my response to Claudio, this is all speculation until Inprise gives some indication as to what the real options are. They have not publicly stated that they are even considering any changes to the status quo.
Joseph Alba: Obviously, a lot of those who object to open sourcing do not really understand what open source means. It simply means that the entire structure and design of the software (or system) is all out in the open and there is nothing to hide. The owner still controls the development path of the product, but because nothing is hidden, bugs get squashed earlier (because everybody gets to participate in the discussion and make suggestions), optimization ideas abound from all over, and the system itself as a whole eventually becomes more robust.
Open sourcing was a very powerful idea which in fact "made" a lot of companies.
Unix, which eventually became the de-facto Operating System out of the many that started with it, became the dominant OS and the robust OS that it is because, believe it or not, it is open source. (Who said Linux is the Open Source Unix... Unix is the open source unix!)
Let me quote Tim Parker:(the TCP/IP guru)
"UNIX is a classic example of an open software platform. UNIX has been around for 30 years. The source code for the UNIX operating system was made available to anyone who wanted it, almost from the start. UNIX's source code is well understood and easy to work with, the result of 30 years of development and improvement. UNIX can be ported to run on practically any hardware platform, eliminating all proprietary dependencies. The attraction of UNIX is not the operating system's features themselves but simply that a UNIX user can run software from other UNIX platforms, that files are compatible from one UNIX system to another (except for disk formats), and that a wide variety of vendors sell products for UNIX."
In its days where OS commanded the top price, an Open Source OS was a strange idea. But, UNIX proved this idea to be a very strong one and can provide great great benefits to its owners.
I think Open Source will in fact save Interbase and make it an even stronger and more robust product in the next 30 years. (Just imagine how much faster IBO can go if Jason can peer into the innards of Interbase. And maybe he can even spot and squash some bugs, and suggest optimizations. But Inprise still controls the development direction of Interbase).
Tony Blomfield: Open Sourcing would have no credibility with my clients at all. I am sorry, but I do not intend to be the missionary carrying the Open Source message to them. I do not believe in Open Source at all. You quote Unix as an example of open source... well if Unix had been properly looked after, we would never have had to tolerate DOS, i.e. the Personal OS of the 80'6 would have been a proper OS based on Unix.
Joseph Alba: Wrong premise.The hardware available then were not suited for Unix. But you are now seeing a comeback by Unix through Linux and Solaris, and it has been proving itself as superior in terms of performance and reliability to Windows NT. But at that time, the mini-computers, mainframes, and even supercomputers had mostly Unix as the operating system.
Tony B: Lets face it, our customers want a product that works, not something that is continually in development by a bunch of geeks on the web - and that is just how open source is perceived by the business community.
Either IB is owned by a company DEDICATED in all aspects of its future including Sales, Development, and Marketing, or it will cease to exist.
Joseph Alba: Again wrong impression of Open Source. Development in Open Source is not continuous. It follows exactly the same trend as the non-Open Source counterparts.
But again, because the source is out there, all of us can participate in putting ideas for all sorts of things like debugging, optimizing, etc.... For example, if IB had been Open Source, I don't think that the Sweep Bug affecting IB 5 could have taken that long to debug. There are still a lot of puzzling things that occasionally happen with IB. Like in Linux, there are the reported lost connections. Some say its because of a too large setting of some buffer in IB. But with Open Source, bugs like this can immediately be fixed. And if your work is so critical, you don't have to wait for the next version. You can fix it yourself, there and then. So, you are not at the mercy of the software's owner.
MIDAS is open source, and even now, bugs are being fixed, and the third parties who are active in Midas development sometimes wonder allowed if Inprise is conscious of their bug fixes already. But usually, Inprise does come around and incorporates those bug fixes.
So, Open Source saves Inprise a ton of money, and it also gains them a world-wide set of "FREE" developers.
Claudio Valderrama: If you could get a profile of the IB developers, I don't believe you will find only one type of person, of course. There are VARs, there're trainers, there're database developers and there're programmers and SW engineers, among others. If you wonder why I separate DB developers from programmers, here's my personal thought: in great systems, one or more people can concentrate in the db stuff without the need to be great Delphi or C/C++ experts. These people can write the db logic, sp, triggers, access arbitration, database design, etc.
My point is if we really have "N" Interbase enthusiasts, how much people are able AND are willing to roll their sleeves and dirty their hands to maintain Open Source IB? Being an independent developer/consultant myself, sometimes I've been contacted to help tracking bugs and correct or optimize code, mainly C++ code. When I'm writing my own applications in Delphi, the last thing I want to do is to step into the database source or the source code for the package I purchased, specially if I'm in a hurry. I've done it sometimes, but it's not my business. The only exception is the VCL itself that I've traced several times for hours.
Joseph Alba wrote:
>Open sourcing was a very powerful idea which in fact "made" a lot of
I do not think Open Source is a bad thing. I only wonder if IB can work with this model because it wasn't born with it and still it's strictly traditional-commercial SW. As far as I have been able to follow Nestcape's evolution, Open Source didn't work as well as expected.
Joseph Alba wrote:
>Unix, which eventually became the de-facto Operating System out of the many
>that started with it, became the dominant OS and the robust OS that it is
>because, believe it or not, it is open source. (Who said Linux is the Open
>Source Unix... Unix is the open source unix!)
Ok, but UNIX started in this way. And let me say one of the harpoons MS has to threaten people going to UNIX/LINUX from Windows is "you are going to a jungle of different, semi-compatible operating systems where the portability promise is broken". MS may be not brilliant doing SW but they have brilliant marketing people that know how to show MS as the only option for your life. When I worked on UNIX, I learned OpenWindows and XWindows were not compatible and there was even another graphic environment proposed by the other companies (DCE or I'm wrong). These days I cannot convince people to go back to command-prompt, so I need a graphical environment, but if there're several flavors of these "windowed" facilities in UNIX, I'm in the same or worse problem as with the nasty differences between Win9X and WinNT in some API calls. And Linux has at least KDE and Gnome. Who can standarize them? Who has the power to make them compatible. Source code, free creativity, you cannot force all people to follow the same trend. If the same happens with Open Source IB, we are lost IMHO.
Joseph Alba wrote:
>I think Open Source will in fact save Interbase and make it an even
stronger >and more robust product in the next 30 years [snip]
Only if we can assure some "implicit obedience" from several groups to an organization that accepts proposals and gives direction to follow. This would ensure coherent versions on several platforms.
Ultimately, this is good if each developer wanting to cooperate is able to make her/his contributions in the platform he/she has the best experience and knowledge, including compilers. On the other side, I wonder how many skilled NLM developers exist as to maintain IB-Netware, for example. I try to guess the first open source version would be released on Linux if IB goes to the this SW model.
However, at risk of being boring, I will repeat that Inprise should make efforts to finish IB6 at least in the platforms that were available in the beta CD-ROMs. This won't take them more than three months more IMHO because IB6 was planned to some date in the first months of 2000 I think. If IB6 is left at this moment, it will take months to independent developers to get the level of knowledge they need to be able to make the final modifications and deliver the product. At least if Borland will give us all the control and the burden of the source code, they must try to put IB6 on the market and then give up.
Ramon van Alteren: Open sourcing is a valid software development technique. It has its advantages and disadvantages, but.......
It's not the answer to anything and requires great commitment from the "controling" or managing entity (Inprise or whoever). If this is not carefully planned and implemented it will fail as will any other development method.....
Since development of new functionality and bug-fixing largely depends on the developer-community in this model I fear that cross-platform development is going to be an issue.
I think, although I don't know this for a fact, that there are less people out there using the Novell version of Interbase than there are Wintel- and Unix users. I fear that this will impair the development of new functionality of the Novell version of Interbase.
I think that all succesfull open-source projects have been open-source from the start of their life or at least very early in their lifecycle. To change from a closed source, closed development method to open source, open development sounds extremely difficult to me. And I am not aware of any succesful projects so far which have been able to make this step. (I admit I haven't looked at Netscape/Mozilla recently)
I think that Inprise is thinking about open-source along lines which have something to do with cutting development costs and I'm almost certain that when starting out with this perspective the switch to open-source will fail. Actually to make this work I think they will have to invest in the product, create infrastructure, document source, train people, provide support etc. etc. And I fear that these will be starting investments that Inprise is not willing to make and IMHO they are essential to make Interbase "tick" as an open source project.
Dalton Calford: A contributor wrote:
>From my experience, and the private responses from those who have contacted me on these issues,
> open sourcing Interbase can be both a boon and a cause of problems.
>From what my research has produced so far is that the only way a open source
>plan will work, is to hire the key individuals from Interbase.
That would mean creating a new company, setting it up as a service support company (very much like Redhat has done with linux) and charging support contracts for Interbase.
This would include the setting up of a single CVS tree with mirrors in various countries. It would include a bug tracking facility and perhaps, if the demand is strong enough, a dedicated distribution of Linux that is specifacally tailored to provide the best Interbase platform.
Such a entity would be both viable and very interesting to the investment community.
If Inprise goes open source, then such an entity could be up and running within months of the announcement.
If Inprise decides to sell the rights to Interbase, then the question of open source is still up for review(but is not a foregone conclusion, it would be up to the investors in the buying group to make such a decision).
The big question about open source is not seeing the source code, it is the license used. If Inprise uses a license similar to Suns licence, then the product is dead. Suns license kills any third party initiative to support thier code.
To clarify, Delphi's VCL is open source, but it is not free and all rights belong to Inprise. This has lead to alot of wonderful addons and tools for Delphi that has allowed it to have so many dedicated developers. People extend Delphi all the time with the knowledge gained from having the source. It also allows for better debuging of your code.
But, due to the license, Delphi has not been ported to any other platform. Even though the code is freely available, it is not free and any porting project would have to start from scratch.
The lazarus project using the fpk compiler, is a fully open GPL'd clone of Delphi. It is still work in progress but the amount of progress over the past few months in amazing (they probly will have a finished product before Inprise does). The base compiler is going through final beta bug fixes and many of the vcl components are already finished and in the lcl. The big difference is, while the main developers are working on linux and windows, other teams are working on OS2, Solaris, HP-UX ports and bringing any bug-fixes back to the main team. All are working together yet separite. The end result is cross-platform, cross-processor (not just intel based) and very powerful.
If delphi's code was truely free, we would already see Delphi on all the platforms.
In contrast, we see Sun steal code from the blackdown team and claim it as thier own. Thier version of open source allows this.
Depending on what form of open source Inprise follows may mean either the begginings of the best SQL product on all platforms or, the end of the product.
[This article may be continued]
Open-Sourcing Interbase - Ed Popkov