Wonderful theory, but what happens when your disk-hog database has to be backed up? How about when it's spanning network servers whose content dwarfs your local index?
You are also forgetting that more disk data means more disk I/Os, and those are not getting appreciably faster if you're going to suck back storage at the rates you're talking about.
I'd also like to know what the point of buying newer, faster hardware is if all that happens is the new machine runs slower because some lazy "programmer" decided efficiency doesn't matter.
It's the vendor/team's job to make the machine usable, not the user's responsibility to keep buying upgrades to support crappy software.
Unfortunately there are too many people who aren't cluing in that we need those cycles to deal with the volume of REAL DATA, not indexing garbage that belongs on a directory/tag search server. Current real data in enterprise and personal data stores is growing far, far faster than the ability of systems to handle the sheer volume.
How are these stats counted? Is it total number of users? Total gigabytes? Total number of hosts? VMs? Servers? Clusters? Enterprise contracts? Individual unit sales?
I know which markets use Sybase, Oracle, and DB/2 and I don't think most of them are feeling much threat at all from SQL Server. If nothing else, all three of the real major players scale on hardware that can crush SQL Server with a single node, never mind the whole cluster.
Actually being more from the POSIX side of things, my first thoughts were NDS or LDAP as directory services to locate the resources. A combination of Kerberos and SSL file encryption/cert keys would even allow your "home" directory to wander around the net with you.
Or you could just "partition" your "home" directory so that your LDAP queries point to the nearest server that knows how to reference your distributed files and resources.
Note: resources.
LDAP can just as easily locate music files, vidcam feeds, or anything else that you can subsequently access via URL. Including object stores, database XML wrapper services, CORBA services, J2EE services, etc.
You could probably even use Apache's negotiation plugin points to select a "best available" connection to the target resource. MQ/CORBA/J2EE if in-network, HTTPS/SOAP/XML if you're out of network, or appropriate resource-specific protocol (e.g. streaming media.)
Mozilla and it's relatives can be "coerced" into treating some of those web-accessible resources as typed document links just by making sure the appropriate mime header information is passed back to a fully-configured client browser. As that could potentially include XUL or XForms interfaces, it's actually a fair bit more powerful concept than WinFS on it's own would be.
I don't remember -- did DCE DFS do XA-compliant network writes, or just Kerberos authorization?
True, but you can do something like use XML flat files that appear as virtual "tables" to the RDBMS, like the object type plugins with PostgreSQL or a Java object type in a commercial database.
That way you get the higher level flexibility while maintaining the lower-level readability/maintainability. All you really need to do is ensure that the XA object interfaces are using the same logic as any "vipw" or configuration GUI's.
Actually I guess you wouldn't even really need to change the format of the underlying config files at all, if you could just get everything sharing an edit lock strategy while you worked on a "better" solution in the future. That way all the existing code that just reads config files doesn't have to change.
Or do you mean the registry settings that point your tools to that directory?
Oops, forgot about your preferences and settings directory that has those minor little security keys and passwords you needed to use those documents.
Keeping things in one place or hierarchy makes sense -- except when it doesn't. Like when they aren't "My" documents, but the departmental budget projections, or the site provisioning schedule, etc.
Personally I find a home "work" directory of symbolic links to my various projects, NFS mounts, etc. gets things pulled together for easy reference, but also lets me remap those components on different systems or workspaces.
It's almost like virtual desktops, saving tab-groups of JSP bookmarks in Mozilla, VFS, or NIS/YP maps maintained by a system admin who actually thinks about how his servers and users actually access the data.
You'd be real impressed to see how useful those approaches are when coordinating a cluster that's backed by a storage solution like EMC, IBM, or Hitachi/HP.
You see, part of the problem is scope. WinFS is from the mentality of a user, the existing solutions are from the mentality of securing and managing entire clusters and enterprise environments. The sheer volume of data makes solutions used by file-system specific approaches collapse as soon as you hit the network.
If you want a parallel, imagine trying to configure an Oracle RAC, IBM Federation cluster, or a Sybase Enterprise cluster to handle hundreds of nodes integrated as a single view. That's what WinFS would have to be able to do before it could be deployed, and it's just not going to get there by focusing on tying it to a file system.
The problem is valid for some users, but the solution is at the fundamentally wrong layer of the system.
I think you make good points and it would be interesting to see someone try to make it really work cross-platform and IP clean, maybe tie it in to some of the work being done to redo system config files with XML so that you end up with an object interface for the system configuration.
LSB would then be kind of an LDAP standard for locating those configuration objects, you could even layer in the Kerberos hooks to provide a real distributed management facility for a cluster or subnet.
As to today and future, that's really the problem. Industry has issues that need to be addressed now, not at some indeterminate future point. It's like Itanium -- great theoretical future benefits, but how do you solve current needs? Where's the short-term payback to justify the vision?
If the patents are truly defensive, they could easily be assigned to a neutral industry party whose charter specifically forbids using any of the patents in IP battles. It seems many other vendors have donated patents to the OSI/FSF/GNU or just blatantly posted public permission to use the patent material.
Propaganda, FUD, and talk are cheap -- they need to start doing.
"Microsoft, OTOH, is in the unique position to implement such an idea..."
Right, like Hans Reiser was in any way impeded from creating his ReiserFS, or IBM was stopped from developing JFS, or SGI was totally impeded in their creation of XFS. And heaven forbid someone should take the source and actually try adding the functionality into the kernel.
If anyone actually wanted WinFS functionality impacting general systems performance, it would have been created by now, at least as a prototype by some dedicated coder. All I see in WinFS is marketing fluff and FUD for the masses, but no actual unique technical need that doesn't already have perfectly serviceable solutions.
Glomming two related services into one blob of unmaintainable code is not necessarily a benefit. A database mapping has the advantage of being able to catalog distributed file systems, including those which don't have any object tag extensions.
The other problem is that it's not uncommon in the database world to spend far more disk indexing complex data for access than it actually takes to store the raw information itself. Do you really want the possibility that your inseperable all-in-one file system is using more space for the equivalent of directory entries than for data itself?
Remember this isn't about special cases like a user too lazy to sort their home directory or documents folder, but applying that overhead to the entire system. With all the tweaks people do to improve general FS performance and reliability, why would anyone think adding overhead is a good idea unless you need, and I mean need those features?
If you do indeed need those features so badly, why not just buy or use one of dozens of existing document storage and search facilities?
WinFS was just trying to find a way to make people think the two ideas were inextricably bound together and in some way unique to Windows. In truth that honour goes to hundreds of document database and repository products and the long-toothed AS400 (or so my cohorts tell me that work on the platform.)
I don't think industry would tolerate a company being that powerful. It might start out independant to develop and prove a conceptual model, but it would pretty much have to shift over to a JCP style process with an industry management board to work in the long run.
In other words, we really need the JCP style bodies to manage standards from a needs based perspective, not a pure technical or vendor lock-in approach. I think Sun/Java have done an exemplary job of implementing that model, though there will always be disgruntled members who want their features merged sooner than they're currently scheduled.
For example, I've been waiting for years for unsigned type support to deal with XML and external systems interfaces more effectively, but it's obvious the JCP has larger issues they're addressing first. No point griping too much about it, just periodically posting a reminder that industry really does need those extra data types for Java to reach it's full potential. It'll happen -- it just takes time and enough demand from the JCP membership to get it scheduled.
Many people like to use the excuse that such patents are for "self defense".
It's also said the best defense is a good offense, so I really don't trust any company that keeps trying to patent the obvious. Instead of wasting money "defending" against bogus patents, how about investing in fixing the patent system?
Maybe I'm phrasing things a bit harshly. Most of the products CA supports and maintains would be abandonware if CA weren't willing to contract the support for business customers who need those products. Just because a product's market wasn't big enough to maintain long-term company viability doesn't mean the product was useless to the market share it did win.
I'll have to look into whether Ingres has full XA support and a cross-platform POSIX support base, but if it runs anywhere SuSE and RedHat do, as well as the commercial *nix vendors, then it could be a viable application database.
The catch is ensuring that customers using that database are providing CA enough of a support and maintenance revenue stream to ensure that OSS Ingres is kept current. With the commercial vendors, it's clear that they have the revenue stream needed to fund the maintenance and support of a core business service. With OSS, you need to make sure the product has the support you need to leverage it in your business infrastructure.
Theoretically if OSS Ingres has support contracts from CA, and PostgreSQL has support contracts from RedHat, then it's up to industry to decide whether those are viable, reliable, secure providers for running their business.
Even Windows is viable to many businesses who trade off available skillsets, price, availability, reliability, uptime, scalability, etc. Every business has different needs, so it's up to them to decide which providers have the best fit and service options to handle their requirements.
At issue are the "leech" customers who think everything should be free as in beer. That needs to be discouraged, maybe by something so simple as making it the responsibility of business to identify who is handling the service and support of their infrastructure. Most likely that's something that affects business insurance pricing and availability rather than being a real threat to the viability of product support services.
No one has ever said that OSS has no cost -- it's just a means of providing the end-customer assurances about the long-term viability of a solution (they have the source), the maintenance of the solution (everyone can vet the code for security gaps), and of allowing different organizations in different areas to provide regionalized support for their local business culture.
I'm not sure about the current releases, but back around 6.3-6.4, Ingres had more features, tunability, and advanced concepts than Oracle, Sybase, or DB/2. It really was a nice database, and the people with their offices in California were pretty sharp.
The problem is that I can't imagine it having grown all that much since then. Once it got taken over by CA, it became a feature for supporting their business, which means it didn't get the focus on enhancements and capability growth that it might have otherwise.
That's really the big problem with anything CA buys. They use it to build their business and service their customers, but they stop doing much (if any) new feature development on the tools they've bought out.
It's amazing how many computing "problems" can be solved by using existing industry standards instead of creating lock-in prototocols and licensing. In the end most businesses prefer open standards where they can leverage competition between implementations.
With the number of applications for simple security wrappers on various media content for delivery, it's clear that the attempts to "patent" the idea of any form of content-specific data delivery is silly. You need a security envelope, a transport or media, and a secure playback facility.
The rest is just competing on the details of quality, reliability, and price as perceived by the customer, not by the RIAA/MPAA or other media manager.
As with most 1.0 products, the intent is to determine if there is enough early adopter interest to continue the development of the planned features and functionality. It's also an opportunity to verify that your target issues and solution are indeed fitting the needs of your market.
1.1 is typically a first cut at addressing the most important functionality features for early adopters to begin generating a decent ROI from the product or service.
The 2.x series usually fleshes out the key components and features that define the long-term vision. Additional releases are primarily to implement that long term vision, keep software up to date with third-party components, and improve security, stability, performance, and scalability over time.
Some products like MQ series have spent their decades of existing fulfilling that vision of a cross-platform reliable messaging backbone. Others like OpenLDAP are pretty simple and take much less time to become relatively stable solutions for specific needs.
Stable core services still leaves plenty of room for adaptation. Consider the accounting industry. Most of their software is implementing the core regulatory requirements for them to achieve and maintain market acceptance. That is a "backbone" requirement for their industry. On top of that they layer their own user interface styles, which some users love and others hate. The same goes for reporting, import/export, connections to online banking facilities, etc. Most core services have thousands of potential applications to explore once you have a reliable processing infrastructure to build on.
The heavy use of anonymous pointers, multi-function entry points, and DLL initialization/release interactions create an absolute nightmare to maintain.
Even for a relatively small project, you have to spend a fair amount of time just getting code separated into mainline and DLL. Then you get the joy of dealing with the weirdities of the Windows variation on process interaction with DLLs.
I can't imagine any way of securing that spaghetti except to scrap the Win32 API and make the.Net framework the Windows programming layer. Then you can get rid of those holdover APIs from DOS-thunker days and replace the kernel with one that was designed for multi-user security.
You can be grateful Microsoft is finally taking security seriously if you like. I look back on 10-15 years of pager calls, system recoveries, and late projects because of bugs, many of which have never been fixed. My patience with their problems and excuses ended a long, long time ago.
Don't forget Microsoft has been around almost exactly as long as GNU.org. Linux is a pup compared to Windows, yet look how much faster that team addresses problems than the much larger team at Microsoft.
If Microsoft's market share begins hurting because of their security issues, they've no one else to blame but themselves. If the industry demands POSIX server APIs and Windows can't deliver, Microsoft has no one to blame but themselves -- the Cygwin team seems to have managed the task.
Microsoft and a lot of other companies need to get back to re-verifying their core business and refocus on producing marketable products and services. Times change, and last decade's sure winner is last year's end-of-life product. A little less focus on the stock market, and a little more on realistic business models and long-term viability.
Right. That's why SP2 came out on time and with so few problems. Not only was it late, it came with new security problems.
I think Bill is just desperate to keep the press from noticing articles like this little tidbit at Newsforge.
As interesting some of the planned features are, they are still dancing around the most important issue: security and timely fixes.
Surely you can't be so naive as to let some FUD like a script utility distract you from the fact the security problems and perpetual scheduling delays!
The problem is not assurance or maintenance plans, the problem is the vendor. IBM DB/2, Oracle, Sybase, AIX, Solaris, HP-UX, SuSE, Mandrake, Red Hat, and hundreds of other vendors provide support and maintenance contracts.
The difference is that they actually fix bugs and security problems instead of pushing "new features".
As to the announced date -- it's from Microsoft. Since when has Microsoft delivered a project on schedule? NT 3.5 was late, 4.0 was late, Win2K/NT5 was late, and so was WinXP.
The only thing I can schedule by the 2006 date is knowing I have at least until then before I need to worry about budget planning. I have work to do that needs to be built with reliable products available now, not more vapourware.
The RIAA has a Canadian branch, they just do their barratry and harassment through that division.
So far they've managed to steal about $30-50 from my pocket for CDRs through their levies on blanks. The fact that I use them primarily for data backup is irrelevant -- they forced the Canadian government to dip into all our pockets for the "piracy losses" of their failing business model.
Nah, the pirates are just like regular users -- they just never delete anything, or keep doing drag-copy instead of moving the files. There are probably entire directories that look like:
john_denvered-asscrofts_boogie.mp3
john_denvered-asscrofts_boogie.mp31
john_denvered-asscrofts_boogie.mp32
john_denvered-asscrofts_boogie.mp33
john_denvered-asscrofts_boogie.mp34...
CICS is just a transaction processor, much as Encina or Tuxedo. There never really has been a huge need for those systems because of their complexity, so there never were that many players.
MQ is in much the same boat. In order to provide cross-enterprise solutions, there has to be a standard for reliable message delivery. There isn't really much to the core code of such a system, so it's more a question of vendor reliability and pricing.
In other words, certain component services for building software are natural monopolies provided they maintain quality, reliability, and keep prices at a point that is reasonable for what they provide.
Java is a bit different. I agree that most standards bodies like ISO or ANSI take too long to come up with standards that include too many hard-to-implement esoteric features that very few people actually need. A community based process like the JCP keeps things under control like a standards body, but allows industry to steer it more effectively to provide the facilities the business customer needs.
For the most part, business wants services while the consumer wants products. Upgrades are like products -- a fixed price purchase that has specific options and features. Services like a weekly stream of minor patches to keep systems secure fit business better. The ability to adapt to changing rules, regulations, and market demand is as important as information security to modern business, especially the P2B/B2B internet segment.
That rounds back to why the commercial vendors will continue exist for the forseeable future -- business needs someone to contract and take legal responsibility for maintaining and supporting the core tools used to run the business.
Sometimes it's not all about price or about technical capability. Sometimes it boils down to estabilishing a verifiable chain of trust to show the customer and your potential market partners that you take your security and infrastructure as seriously as they take theirs. If they don't, then you are dealing with risk-benefit calculations, essentially gambling with the survival of the business.
I don't like gambling with business. It's bad for the revenue stream, for the customers, for the staff, and puts you at risk of losing everything to a coin toss. Some may find that thrilling, I just find it irresponsible.
I'm not frightened because I honestly don't see industry tolerating Microsoft's continued waste of other company's resources.
For all the congress and law that Microsoft may have bought, they still don't employ that many people on the national scale. Microsoft threatens to move to Vancouver (for example), and some revenue and jobs are gone.
IBM, Sun, HP, etc. all make that same threat (as they already have international offices), and suddenly it's the entire industry that is telling the government "fix this, or else."
When it comes down to the line, I have no doubt that Microsoft will be slapped down for the simple reason that if they're not, everyone else has to quit so that Microsoft can own the world. Ain't gonna happen.
Not at all. GNU was started in response to the lockin of IBM, AT&T, DEC, Pyramid, and others. Microsoft wasn't much more than Bill Gates' brain-fart at the point GNU began.
In fact, Microsoft began their approach to marketting from the same perspective as GNU: freedom from the big vendors and outrageous licensing fees.
What both were really saying is that competition is good for business. It leads to better solutions.
What Microsoft forgot is that business only benefits when the better solutions are generic and can be applied to a variety of problems without fear of barratry.
They never have been very good at delivering quality -- the first release of their C compiler didn't even handle pointers properly, making it effectively a useless, overpriced prototype of a compiler at best. The combination of market domination tactics and refusal to acknowledge or correct defects in their products or their support of industry standards is why Microsoft stands alone.
Despite their size, Microsoft is still only a small percentage of the global IT market. Outside the US, they don't matter half as much as they'd like. China, India, Japan, Europe, Australia, the UK -- most regions are backing the same international standards as the big vendors.
Standards level the playing field. It makes it easier to port apps, it makes it easier to implement secure systems because everyone is familiar with the essential capabilities and how to use them. With Windows, it's a perpetual churn as Microsoft tries to come up with their own way of doing what the industry has already been doing for years.
It seems a tremendous waste of energy and resources that costs society and industry an awful lot of overtime, lost weekends, lost data, and money. All to service the collective egos of one company's board of directors.
It just baffles me that anyone thinks Microsoft has real power when push comes to shove. When the final line is drawn, on one side are not just the standards, but everyone that Microsoft has pissed off by losing data, overcharging for an "upgrade" that still doesn't fix fundamental problems, etc.
More importantly, it is the revenue streams and survival of global industry vs. one company.
Slashdot Mirror
Wonderful theory, but what happens when your disk-hog database has to be backed up? How about when it's spanning network servers whose content dwarfs your local index?
You are also forgetting that more disk data means more disk I/Os, and those are not getting appreciably faster if you're going to suck back storage at the rates you're talking about.
I'd also like to know what the point of buying newer, faster hardware is if all that happens is the new machine runs slower because some lazy "programmer" decided efficiency doesn't matter.
It's the vendor/team's job to make the machine usable, not the user's responsibility to keep buying upgrades to support crappy software.
Unfortunately there are too many people who aren't cluing in that we need those cycles to deal with the volume of REAL DATA, not indexing garbage that belongs on a directory/tag search server. Current real data in enterprise and personal data stores is growing far, far faster than the ability of systems to handle the sheer volume.
How are these stats counted? Is it total number of users? Total gigabytes? Total number of hosts? VMs? Servers? Clusters? Enterprise contracts? Individual unit sales?
I know which markets use Sybase, Oracle, and DB/2 and I don't think most of them are feeling much threat at all from SQL Server. If nothing else, all three of the real major players scale on hardware that can crush SQL Server with a single node, never mind the whole cluster.
Actually being more from the POSIX side of things, my first thoughts were NDS or LDAP as directory services to locate the resources. A combination of Kerberos and SSL file encryption/cert keys would even allow your "home" directory to wander around the net with you.
Or you could just "partition" your "home" directory so that your LDAP queries point to the nearest server that knows how to reference your distributed files and resources.
Note: resources.
LDAP can just as easily locate music files, vidcam feeds, or anything else that you can subsequently access via URL. Including object stores, database XML wrapper services, CORBA services, J2EE services, etc.
You could probably even use Apache's negotiation plugin points to select a "best available" connection to the target resource. MQ/CORBA/J2EE if in-network, HTTPS/SOAP/XML if you're out of network, or appropriate resource-specific protocol (e.g. streaming media.)
Mozilla and it's relatives can be "coerced" into treating some of those web-accessible resources as typed document links just by making sure the appropriate mime header information is passed back to a fully-configured client browser. As that could potentially include XUL or XForms interfaces, it's actually a fair bit more powerful concept than WinFS on it's own would be.
I don't remember -- did DCE DFS do XA-compliant network writes, or just Kerberos authorization?
True, but you can do something like use XML flat files that appear as virtual "tables" to the RDBMS, like the object type plugins with PostgreSQL or a Java object type in a commercial database.
That way you get the higher level flexibility while maintaining the lower-level readability/maintainability. All you really need to do is ensure that the XA object interfaces are using the same logic as any "vipw" or configuration GUI's.
Actually I guess you wouldn't even really need to change the format of the underlying config files at all, if you could just get everything sharing an edit lock strategy while you worked on a "better" solution in the future. That way all the existing code that just reads config files doesn't have to change.
Ahh, you mean the ubiquitous "home" directory.
Or do you mean the registry settings that point your tools to that directory?
Oops, forgot about your preferences and settings directory that has those minor little security keys and passwords you needed to use those documents.
Keeping things in one place or hierarchy makes sense -- except when it doesn't. Like when they aren't "My" documents, but the departmental budget projections, or the site provisioning schedule, etc.
Personally I find a home "work" directory of symbolic links to my various projects, NFS mounts, etc. gets things pulled together for easy reference, but also lets me remap those components on different systems or workspaces.
It's almost like virtual desktops, saving tab-groups of JSP bookmarks in Mozilla, VFS, or NIS/YP maps maintained by a system admin who actually thinks about how his servers and users actually access the data.
You'd be real impressed to see how useful those approaches are when coordinating a cluster that's backed by a storage solution like EMC, IBM, or Hitachi/HP.
You see, part of the problem is scope. WinFS is from the mentality of a user, the existing solutions are from the mentality of securing and managing entire clusters and enterprise environments. The sheer volume of data makes solutions used by file-system specific approaches collapse as soon as you hit the network.
If you want a parallel, imagine trying to configure an Oracle RAC, IBM Federation cluster, or a Sybase Enterprise cluster to handle hundreds of nodes integrated as a single view. That's what WinFS would have to be able to do before it could be deployed, and it's just not going to get there by focusing on tying it to a file system.
The problem is valid for some users, but the solution is at the fundamentally wrong layer of the system.
I think you make good points and it would be interesting to see someone try to make it really work cross-platform and IP clean, maybe tie it in to some of the work being done to redo system config files with XML so that you end up with an object interface for the system configuration.
LSB would then be kind of an LDAP standard for locating those configuration objects, you could even layer in the Kerberos hooks to provide a real distributed management facility for a cluster or subnet.
As to today and future, that's really the problem. Industry has issues that need to be addressed now, not at some indeterminate future point. It's like Itanium -- great theoretical future benefits, but how do you solve current needs? Where's the short-term payback to justify the vision?
If the patents are truly defensive, they could easily be assigned to a neutral industry party whose charter specifically forbids using any of the patents in IP battles. It seems many other vendors have donated patents to the OSI/FSF/GNU or just blatantly posted public permission to use the patent material.
Propaganda, FUD, and talk are cheap -- they need to start doing.
"Microsoft, OTOH, is in the unique position to implement such an idea..."
Right, like Hans Reiser was in any way impeded from creating his ReiserFS, or IBM was stopped from developing JFS, or SGI was totally impeded in their creation of XFS. And heaven forbid someone should take the source and actually try adding the functionality into the kernel.
If anyone actually wanted WinFS functionality impacting general systems performance, it would have been created by now, at least as a prototype by some dedicated coder. All I see in WinFS is marketing fluff and FUD for the masses, but no actual unique technical need that doesn't already have perfectly serviceable solutions.
Glomming two related services into one blob of unmaintainable code is not necessarily a benefit. A database mapping has the advantage of being able to catalog distributed file systems, including those which don't have any object tag extensions.
The other problem is that it's not uncommon in the database world to spend far more disk indexing complex data for access than it actually takes to store the raw information itself. Do you really want the possibility that your inseperable all-in-one file system is using more space for the equivalent of directory entries than for data itself?
Remember this isn't about special cases like a user too lazy to sort their home directory or documents folder, but applying that overhead to the entire system. With all the tweaks people do to improve general FS performance and reliability, why would anyone think adding overhead is a good idea unless you need, and I mean need those features?
If you do indeed need those features so badly, why not just buy or use one of dozens of existing document storage and search facilities?
WinFS was just trying to find a way to make people think the two ideas were inextricably bound together and in some way unique to Windows. In truth that honour goes to hundreds of document database and repository products and the long-toothed AS400 (or so my cohorts tell me that work on the platform.)
I don't think industry would tolerate a company being that powerful. It might start out independant to develop and prove a conceptual model, but it would pretty much have to shift over to a JCP style process with an industry management board to work in the long run.
In other words, we really need the JCP style bodies to manage standards from a needs based perspective, not a pure technical or vendor lock-in approach. I think Sun/Java have done an exemplary job of implementing that model, though there will always be disgruntled members who want their features merged sooner than they're currently scheduled.
For example, I've been waiting for years for unsigned type support to deal with XML and external systems interfaces more effectively, but it's obvious the JCP has larger issues they're addressing first. No point griping too much about it, just periodically posting a reminder that industry really does need those extra data types for Java to reach it's full potential. It'll happen -- it just takes time and enough demand from the JCP membership to get it scheduled.
Many people like to use the excuse that such patents are for "self defense".
It's also said the best defense is a good offense, so I really don't trust any company that keeps trying to patent the obvious. Instead of wasting money "defending" against bogus patents, how about investing in fixing the patent system?
Maybe I'm phrasing things a bit harshly. Most of the products CA supports and maintains would be abandonware if CA weren't willing to contract the support for business customers who need those products. Just because a product's market wasn't big enough to maintain long-term company viability doesn't mean the product was useless to the market share it did win.
I'll have to look into whether Ingres has full XA support and a cross-platform POSIX support base, but if it runs anywhere SuSE and RedHat do, as well as the commercial *nix vendors, then it could be a viable application database.
The catch is ensuring that customers using that database are providing CA enough of a support and maintenance revenue stream to ensure that OSS Ingres is kept current. With the commercial vendors, it's clear that they have the revenue stream needed to fund the maintenance and support of a core business service. With OSS, you need to make sure the product has the support you need to leverage it in your business infrastructure.
Theoretically if OSS Ingres has support contracts from CA, and PostgreSQL has support contracts from RedHat, then it's up to industry to decide whether those are viable, reliable, secure providers for running their business.
Even Windows is viable to many businesses who trade off available skillsets, price, availability, reliability, uptime, scalability, etc. Every business has different needs, so it's up to them to decide which providers have the best fit and service options to handle their requirements.
At issue are the "leech" customers who think everything should be free as in beer. That needs to be discouraged, maybe by something so simple as making it the responsibility of business to identify who is handling the service and support of their infrastructure. Most likely that's something that affects business insurance pricing and availability rather than being a real threat to the viability of product support services.
No one has ever said that OSS has no cost -- it's just a means of providing the end-customer assurances about the long-term viability of a solution (they have the source), the maintenance of the solution (everyone can vet the code for security gaps), and of allowing different organizations in different areas to provide regionalized support for their local business culture.
I'm not sure about the current releases, but back around 6.3-6.4, Ingres had more features, tunability, and advanced concepts than Oracle, Sybase, or DB/2. It really was a nice database, and the people with their offices in California were pretty sharp.
The problem is that I can't imagine it having grown all that much since then. Once it got taken over by CA, it became a feature for supporting their business, which means it didn't get the focus on enhancements and capability growth that it might have otherwise.
That's really the big problem with anything CA buys. They use it to build their business and service their customers, but they stop doing much (if any) new feature development on the tools they've bought out.
It's amazing how many computing "problems" can be solved by using existing industry standards instead of creating lock-in prototocols and licensing. In the end most businesses prefer open standards where they can leverage competition between implementations.
With the number of applications for simple security wrappers on various media content for delivery, it's clear that the attempts to "patent" the idea of any form of content-specific data delivery is silly. You need a security envelope, a transport or media, and a secure playback facility.
The rest is just competing on the details of quality, reliability, and price as perceived by the customer, not by the RIAA/MPAA or other media manager.
Windows 1.0 had virtually no market share.
As with most 1.0 products, the intent is to determine if there is enough early adopter interest to continue the development of the planned features and functionality. It's also an opportunity to verify that your target issues and solution are indeed fitting the needs of your market.
1.1 is typically a first cut at addressing the most important functionality features for early adopters to begin generating a decent ROI from the product or service.
The 2.x series usually fleshes out the key components and features that define the long-term vision. Additional releases are primarily to implement that long term vision, keep software up to date with third-party components, and improve security, stability, performance, and scalability over time.
Some products like MQ series have spent their decades of existing fulfilling that vision of a cross-platform reliable messaging backbone. Others like OpenLDAP are pretty simple and take much less time to become relatively stable solutions for specific needs.
Stable core services still leaves plenty of room for adaptation. Consider the accounting industry. Most of their software is implementing the core regulatory requirements for them to achieve and maintain market acceptance. That is a "backbone" requirement for their industry. On top of that they layer their own user interface styles, which some users love and others hate. The same goes for reporting, import/export, connections to online banking facilities, etc. Most core services have thousands of potential applications to explore once you have a reliable processing infrastructure to build on.
The heavy use of anonymous pointers, multi-function entry points, and DLL initialization/release interactions create an absolute nightmare to maintain.
Even for a relatively small project, you have to spend a fair amount of time just getting code separated into mainline and DLL. Then you get the joy of dealing with the weirdities of the Windows variation on process interaction with DLLs.
I can't imagine any way of securing that spaghetti except to scrap the Win32 API and make the .Net framework the Windows programming layer. Then you can get rid of those holdover APIs from DOS-thunker days and replace the kernel with one that was designed for multi-user security.
You can be grateful Microsoft is finally taking security seriously if you like. I look back on 10-15 years of pager calls, system recoveries, and late projects because of bugs, many of which have never been fixed. My patience with their problems and excuses ended a long, long time ago.
Don't forget Microsoft has been around almost exactly as long as GNU.org. Linux is a pup compared to Windows, yet look how much faster that team addresses problems than the much larger team at Microsoft.
If Microsoft's market share begins hurting because of their security issues, they've no one else to blame but themselves. If the industry demands POSIX server APIs and Windows can't deliver, Microsoft has no one to blame but themselves -- the Cygwin team seems to have managed the task.
Microsoft and a lot of other companies need to get back to re-verifying their core business and refocus on producing marketable products and services. Times change, and last decade's sure winner is last year's end-of-life product. A little less focus on the stock market, and a little more on realistic business models and long-term viability.
We will not cut corners on product excellence.
Right. That's why SP2 came out on time and with so few problems. Not only was it late, it came with new security problems.
I think Bill is just desperate to keep the press from noticing articles like this little tidbit at Newsforge.
As interesting some of the planned features are, they are still dancing around the most important issue: security and timely fixes.
Surely you can't be so naive as to let some FUD like a script utility distract you from the fact the security problems and perpetual scheduling delays!
Free downloads?
Good for you.
I don't appreciate paying for your music.
The problem is not assurance or maintenance plans, the problem is the vendor. IBM DB/2, Oracle, Sybase, AIX, Solaris, HP-UX, SuSE, Mandrake, Red Hat, and hundreds of other vendors provide support and maintenance contracts.
The difference is that they actually fix bugs and security problems instead of pushing "new features".
As to the announced date -- it's from Microsoft. Since when has Microsoft delivered a project on schedule? NT 3.5 was late, 4.0 was late, Win2K/NT5 was late, and so was WinXP.
The only thing I can schedule by the 2006 date is knowing I have at least until then before I need to worry about budget planning. I have work to do that needs to be built with reliable products available now, not more vapourware.
The RIAA has a Canadian branch, they just do their barratry and harassment through that division.
So far they've managed to steal about $30-50 from my pocket for CDRs through their levies on blanks. The fact that I use them primarily for data backup is irrelevant -- they forced the Canadian government to dip into all our pockets for the "piracy losses" of their failing business model.
Nah, the pirates are just like regular users -- they just never delete anything, or keep doing drag-copy instead of moving the files. There are probably entire directories that look like: john_denvered-asscrofts_boogie.mp3 john_denvered-asscrofts_boogie.mp31 john_denvered-asscrofts_boogie.mp32 john_denvered-asscrofts_boogie.mp33 john_denvered-asscrofts_boogie.mp34 ...
CICS is just a transaction processor, much as Encina or Tuxedo. There never really has been a huge need for those systems because of their complexity, so there never were that many players.
MQ is in much the same boat. In order to provide cross-enterprise solutions, there has to be a standard for reliable message delivery. There isn't really much to the core code of such a system, so it's more a question of vendor reliability and pricing.
In other words, certain component services for building software are natural monopolies provided they maintain quality, reliability, and keep prices at a point that is reasonable for what they provide.
Java is a bit different. I agree that most standards bodies like ISO or ANSI take too long to come up with standards that include too many hard-to-implement esoteric features that very few people actually need. A community based process like the JCP keeps things under control like a standards body, but allows industry to steer it more effectively to provide the facilities the business customer needs.
For the most part, business wants services while the consumer wants products. Upgrades are like products -- a fixed price purchase that has specific options and features. Services like a weekly stream of minor patches to keep systems secure fit business better. The ability to adapt to changing rules, regulations, and market demand is as important as information security to modern business, especially the P2B/B2B internet segment.
That rounds back to why the commercial vendors will continue exist for the forseeable future -- business needs someone to contract and take legal responsibility for maintaining and supporting the core tools used to run the business.
Sometimes it's not all about price or about technical capability. Sometimes it boils down to estabilishing a verifiable chain of trust to show the customer and your potential market partners that you take your security and infrastructure as seriously as they take theirs. If they don't, then you are dealing with risk-benefit calculations, essentially gambling with the survival of the business.
I don't like gambling with business. It's bad for the revenue stream, for the customers, for the staff, and puts you at risk of losing everything to a coin toss. Some may find that thrilling, I just find it irresponsible.
Once they're convicted, then can we shoot them?
I'm not frightened because I honestly don't see industry tolerating Microsoft's continued waste of other company's resources.
For all the congress and law that Microsoft may have bought, they still don't employ that many people on the national scale. Microsoft threatens to move to Vancouver (for example), and some revenue and jobs are gone.
IBM, Sun, HP, etc. all make that same threat (as they already have international offices), and suddenly it's the entire industry that is telling the government "fix this, or else."
When it comes down to the line, I have no doubt that Microsoft will be slapped down for the simple reason that if they're not, everyone else has to quit so that Microsoft can own the world. Ain't gonna happen.
Not at all. GNU was started in response to the lockin of IBM, AT&T, DEC, Pyramid, and others. Microsoft wasn't much more than Bill Gates' brain-fart at the point GNU began.
In fact, Microsoft began their approach to marketting from the same perspective as GNU: freedom from the big vendors and outrageous licensing fees.
What both were really saying is that competition is good for business. It leads to better solutions.
What Microsoft forgot is that business only benefits when the better solutions are generic and can be applied to a variety of problems without fear of barratry.
They never have been very good at delivering quality -- the first release of their C compiler didn't even handle pointers properly, making it effectively a useless, overpriced prototype of a compiler at best. The combination of market domination tactics and refusal to acknowledge or correct defects in their products or their support of industry standards is why Microsoft stands alone.
Despite their size, Microsoft is still only a small percentage of the global IT market. Outside the US, they don't matter half as much as they'd like. China, India, Japan, Europe, Australia, the UK -- most regions are backing the same international standards as the big vendors.
Standards level the playing field. It makes it easier to port apps, it makes it easier to implement secure systems because everyone is familiar with the essential capabilities and how to use them. With Windows, it's a perpetual churn as Microsoft tries to come up with their own way of doing what the industry has already been doing for years.
It seems a tremendous waste of energy and resources that costs society and industry an awful lot of overtime, lost weekends, lost data, and money. All to service the collective egos of one company's board of directors.
It just baffles me that anyone thinks Microsoft has real power when push comes to shove. When the final line is drawn, on one side are not just the standards, but everyone that Microsoft has pissed off by losing data, overcharging for an "upgrade" that still doesn't fix fundamental problems, etc.
More importantly, it is the revenue streams and survival of global industry vs. one company.