At this year's Expo, the hall for the keynote was moved to the South side of Moscone to accomodate more people. It was laid out for 4,000, with an estimated 85% attendance. On the exhibit floor, there were fewer exhibitors, although the floor space was only slightly smaller. Wider aisles and larger booths made up the difference.
For someone keeping up to date with announcements, there were few surprises. Lighthouse have announced two (!?) spreadsheets and a database, both bought in, and will be selling FrameMaker for Intel. Some big exhibitors from last year were absent, or combined with other stands, and there were few new exhibitors. OTI are allegedly close to collapse, and are seeking an investor or buyer.
According to Steve Jobs (to revert to the keynote), NeXT is now a software company with a mission. That mission is to lead the object revolution and become an accepted alternative to Microsoft in the corporate market. NeXT want to be the largest object company.
NeXT's management team is apparently stable and experienced, with members in place from 8 months to a couple of years, and a grand total of 30 years computer experience among about 8 of them (as an aside, I found this bit hilarious; P & L Systems management team of 2 has more than that between us).
NeXT is now 275 employees, 100 in R & D, 100 in Sales & Marketing, and the rest split between services and G&A. Revenue growth has been consistent, from 2 million in q2 '93, through 4, 5 and 8 to an anticipated 10 million in quarter 2 '94.
NeXT will be profitable every quarter in 1994 (applause); and they now have the right partnerships in place for growth.
Looking at revenue in quarter 1 1994, ParcPlace are the nearest object company, with $8 million revenue. Others totalled are under $4 million. ParcPlace revenue splits into $4.5 million from software, the rest being services. This means that NeXT are already the largest supplier of object software.
Talking about customers, NeXT have now loudly disowned the academic market. For the past three years, strategy has been for mission critical custom apps. This has been the province of mainframe systems, with 18 month plus delivery times. These were used to provide back office systems. NeXT is now being used to deploy apps in the front office with greatly reduced delivery times. For example, MCI used a NeXT system to create the "Family and Friends" campaign. There was no response from competitors for over six months, because they were unable to deploy the custom software needed to create this type of product.
The general strategy involves downsizing, using client/server systems, with applications that are much closer to customer requirements. For each new product or service, the user must create a custom application to drive it.
A few of the customer (these slides went by very quickly): Fannie Mae; Chrysler Financial with 1,250 users and 2,500 by year end. Chrysler wrote their first 5 custom apps in 90 days, and over the next 90 developer another 19 custom apps.
Market segments were financial, communications, healthcare and government. NTT (Nippon Telecom) were one of the bigger new comms customers. In the US government, there are 10,000 seats with over 100 custom apps (which is not what I hear, but never mind; to balance that, I heard of a bid for 25,000 Digital PCs to run NeXTSTEP that was closed this week).
Another category put up was "reengineering", which included Linklaters & Paines, so I am at a loss to work out what it means.
The current sales strategy is for direct sales, supported by the "object channel". These are systems integrators who have paid $7,000 for the privilege of competing with NeXT for sales :^>. Sixty of these have been signed up. The final component is OEMs (HP, Sun, Canon, and PC manufacturers). In 1994, NeXT expect that 75% of sales will come from the direct sales force.
an OS independent object standard.
A tape from Ed Zander reinforced Sun's commitment to OpenStep. There is no alternative strategy for development. Why did they do it? An investment in object technology, the need to develop custom apps, and the fact that OpenStep is an open system with a published standard. Sun have "no object alternatives". Sun ship 350,000 to 400,000 Solaris licenses per year: these will all be shipped with OpenStep.
They represent one third of the Unix market, with $10 billion turnover. OpenStep will run on top of OSF/1. Digital plus Sun plus HP own 68% of the Unix desktop.
OpenStep will be available in draft on June 30 for ftp. The final version is due September 30. The API and trademark will be free for all systems that pass a validation suite.
In '94 NeXT were shipping 3.x; in '96, they will ship Mecca (Steve says: "why stop at Cairo?").
Mecca will be OS independent, with new memory allocation methods, a new foundation library, a new distributed object mechanism, unicode support, and a new API. There will be automated conversion tools to support the new API; conversion is estimated to take less than 1 man/month, more typically around 10 man days.
All versions of OpenStep will ship in '95: NeXT, HP, Digital and Sun.
The object market is a three horse race, between NeXTSTEP, Cairo and Taligent.
NeXT's new products are in Foundation, Interface, Database and Distribution categories. Core products are NeXTSTEP User and Developer.
3.2 shipped in '93 on most platforms. 3.2 PA-RISC will ship at the end of July, for the same price as Intel. 3.3 is due in late '94. Features will be: customer driven changes; PC enablement (this means drivers), scalability, reliability, international support (but not Kanji), and will be a User release only (3.2 developer will still work).
Portables: 8 bit colour, power management, PCMCIA support.
Desktops: PCI bus, and "plug and play".
Mail: MIME, draft messages, hierarchical mailboxes, and custom header support.
Systems Admin: improved tools, net installation, improved netinfo, and support for IP multicast and tunnelling.
Beta will be in August, ship in November. This will come with a new version of SoftPC, which includes enhanced mode support, better netware performance, and bug fixes.
he was using an internal app called PSdemo, which showed some really rapid PS displays. He also used an NEC Versa, with power management. This is an extra preferences module, and adds extra animation to the Preferences icon (battery life).
in '95, 4.0 is due. It will include OpenStep compatability, binary compatability with 3.x apps, Mail 4.0, Kanji, and will still run on black NeXTs. This is scheduled for summer '95.
4.0 will include a new Developer release. This will have a new InterfaceBuilder, new ProjectBuilder, shared library support, and "fix and go". This latter is an improved debug cycle that reduces cycle time to seconds. This is scheduled for summer '95 also.
Also with 4.0 is a new SoftPC. Enhancements will include support for Chicago, rootless windows (so you will have independent Windows windows floating around between NeXTSTEP windows) and will allow you to launch Windows apps from the Workspace.
PDO (Portable Distributed Objects) shipped in November '93 for HP, priced between $5,000 and $10,000. PDO 2.0 will run on SunOS, Solaris, and includes C++ support. The beta will be available this year, and will be a free upgrade. OSF/1 will be included during '94.
The current approach to database support (aka DBKit) allows you to fetch the result of queries into windows objects, with automatic generation of SQL. The user interface is built up in InterfaceBuilder, and aims at database backend independence. PowerBuilder is much the same. The main problem with this approach is that internal program logic that has to be applied to this data can only be applied at the window level, and has to be duplicated for each window. Hence this makes code resuse very hard, and generates code that is difficult to maintain.
The answer to this is the Enterprise Objects Framework (EOF). This works by feeding data from the database not into windows objects, but directly into business objects, making any logic that has to be applied to the data centralised into one place. This lets us use standard approaches such as subclassing objects for faster code reuse. This is currently client only, but there will be future version of EOF that will be server code.
The alpha version went to 24 customers. The beta was distributed to all developer registrations, and the final version will ship in October. Price is $299.
NeXTIME. Ships NOW. There are no hardware requirements. It uses the QuickTime file format, and will play QT files from disk, CDROM, or across a network. It uses the CinePak compressor. Price is $99, but there was a show special for $20. NeXTIME 2.0 will allow you to record on PCs, will have an API, and will allow MPEG. It will ship with 4.0.
Steve ended with a long stream of soundbites about how the industry must cooperate to support the only viable alternative to Microsoft: "if we don't do this, no one else will"; "it is time for companies to ... join with us".
This year the embarrasing sound system crashes didn't happen (NeXT were taking no chances on that); however, NeXTIME crashed enough in demo to make up for it, and the projector was playing up enough to be distracting..
Back from the keynote for a while, the general feeling was good, with some clouds on the horizon. On the release of NeXTIME, several developers said that at last NeXT can get back to working on things that will be useful to everyone else. Getting OpenStep to include new 4.0 features, rather than freezing at the 3.x level is felt to be a good thing, to preserve some of the spirit of revolution that always came with NeXTSTEP. On the down side, it was felt that this approach finally kills off any support for independent software vendors. This feeling wasn't shared by the major ISVs, as it happens.
From general discussions, it was clear that NeXT cooperation with third parties to sell more NeXTSTEP is at an all time low. Every Object Channel partner that I spoke to had a story to tell about how NeXT would compete with them for sales. In particular with ISVs, NeXT don't seem to understand applications. Several stories came out about NeXT insisting on ludicrous price cuts for sale to individual customers, and insisting on distribution agreements with incompetent OEMs and resellers. I could give specific examples, but that wouldn't help. There are a lot of young and naive NeXTSTEP developers, but the best marketing skills that I know anywhere are working in the ISV channel.
There has been a lot of change in the past year, and there is more to come. This will affect both NeXT and the third party market much more than customers, curiously enough. There is absolutely no doubt that NeXT has the only serious OO development environment, and that the direction shown by OpenStep and 4.0 is absolutely correct. The third parties who have bent to keep with the MCCA direction, and who have rigorously resisted profligate spending won't go away anytime soon. The famous NeXT "level playing field", and the now notorious saying: "s**t floats" are bigger hazards than any competing software.