I've been working in software organizations for over 20 years at companies like HP and Sun. Math and Science are not as important as good logic and tenacity! Most of the outsourced work to other countries is not complicated work... sustaining and testing is more about attitude than hard core math and science. That is what's mostly being outsourced.
Sun in the past has not been able to develop as desktop interface that gained any serious traction. Sun's offerings in the past have been Sunview, NeWs, OpenLook, CDE, GNOME. And now a Java 3d desktop...
Sun has little experience dealing with end-users directly. At best I think they'll develop technology that some other entity (i.e IBM, BEA) will drive into the mainsteam.
Sun dumped most of the 3D software engineers a few years back, loosing alot of domain expertise
Not to sure the Linux world cares much about a Java Desktop, so Sun will cough up 99% of the engineering resources to build this stuff
Seems to me new applications would be a significantly higher priority
Hey... you are not far off target with #4... The two most popular scripting languages at Sun used by the Solaris operating system organization is bourne shell and ksh... when I attempted to write to tools using perl I was told to use either sh or ksh... the computing environment at Sun in the solaris organization is based on 1980's technologies and tools... its really sad... basically they've built so much on-top of bad processes that everyone that wants to make a change, makes the change on top of the existing mess using outdated tools... And don't even get me going on the Java organization... its so confuscated people can't get their work done...
Last year I was on a project that required 17 Java programmers to complete a 3 month project that could have been completed by 1 PHP programmer and 1 systems programmer in 30 days... why would the company proceed in this manner. 1) its like a new tool-toy, its used for everything even when it should not be. 2) its politically correct. 3) its all part of the marketing campaign that must be one of the most successful in hi-tech...
Hmmm... so as I best recall from working at places like HP and Sun for most of the 80's and 90's...
Unix Kernel developers were king of the hill. I'm talking about the folks that developed the VM system, Threading, Networking, core kernel work... (i.e not so much device drivers as lots on none core kernel folks contributed)
Thru the late 80's into the mid 90's the graphics developers i.e hardware/software think framebuffers/opengl become very much in demand, required alot of math besides programming
Core database developers have (i.e oracle kernel, sybase kernel) been hot the entire time
In regards to OS Development vs IT careers there are alot of variables. They are such different disciplines and a very different culture. I've worked in both, started in IT (4yrs) then moved into engineering; database, networking, os, graphics (17 years). Finding both a culture and type of work can be challengine but very rewarding once you've found what meets your requirements. In regards to compensation, it really is supply and demand that drives the salary ranges.
So in a long winded way... how to get started in OS development... At places like HP and Sun there are multiple tracks to doing this
* Customer Support (in-house) people are assigned to specific areas of system functionality, i.e system commands, system libraries, networking, device drivers etc...
* Sustaining Teams, they are supporting the customer support folks provide patches for released baselines
* Technical Marketing, They are working closely with development and customers
* Consulting Teams, They too are working closely with development and high profile customers
* Any development team that is developming new hardware for an OS platform (graphics, storage, et c...)
* An OS development organization... just get your foot in the door and the rest will happen as you develop your skills. You might consider (unit test development, system services development, buildmeister, lab administrator)
My experiences and observations based on working at Sun thru most of the 90's and recently on a contract is just the opposite. As people move up the corporate latter they become more conservative as there is more at risk, more to loose. Decision making takes longer and conflicts go unresolved because of the increased risk. A pattern of Good News up, Bad News down becomes the norm. And with the Peter Principle in full affect, accountability goes at the window.
-psk
Re:It Aint The Managers Fault or the Corporations
on
The Career Programmer
·
· Score: 1
I wish I could... I don't know of any... Lots of great books have been written on Software Design and Methodologies. I'm not aware of any decent books on 'software infrastructure to support the software life cycle'. I'm inclined to think if anything of significance is going to happen in this area, it will be in open source. In regards to suggesting books... The classics by Pressman, Yourdan, DiMarco are good, so are the more recent books on xtreme-programming and also the books from MS (debuging the development cycle).
But in terms of building a high performance software development organization... Nada
-psk
It Aint The Managers Fault or the Corporations
on
The Career Programmer
·
· Score: 1, Interesting
I've been working in the computer industry since 1978, which included long stints at HP and Sun in the HP-UX and Solaris Networking/Operating Systems Organizations
From my perspective the scenarios that are dicussed in this thread and many others are due to both the flexibility of software development and the rigidity required to develop the desired software, resulting in a continuous conflict that few teams are ever able to resolve.
The processes and practices used by most companies in silicon valley are still stuck in the 1980's. The infratructures used to support development efforts at most companies are also stuck in the 1980's. Case in point: Most popular source control tools are based on SCCS, RCS and ClearCase (think Apollo DSEE from the 80's). Most popular editors are based on Emacs, Vi. Most popular IDEs are based on Microsoft and Borland from the 1980s. Most user interfaces are based on 1970's and 1980's look and feel (think HTML forms, think Microsoft, think Apple)
One other key issues I've observed in silicon valley when it comes to software development: We have architects of technology but we don't have architects of product. To test this out in your situation, try to determine who owns the software development lifecycle and who owns the product.
BigusKramicus
The person writing this article has the right idea in terms of software bugs, software complexity... but does appear to have a good understanding of what it takes to change the culture of a software organization let alone the software industry.
In the commercial software business it starts with requirements management (which hardly exists) and Design. Design i.e architecture as a discipline ignores products and focuses on technology. Go to any software company in Silicon Valley and you will not be able to identify who owns product architecture, only who owns product technology. This results in a really interesting scenario where no one in the company really owns the product (i.e accountability). Since the product was never defined up-front and not owned by anyone, the quality is always going to be mediocre at best.
If we take a quick stab at what the definition of product is: 1) application or system, 2) installation, upgrades), 3) administration, 4) documetation, 5) training, 6) customer support, 7) integration and consulting
I've had the opportunity to work at HP and Sun in HPUX/SunOS/Solaris organizations and know first hand why quality is a problem... Its never been a priority. The way you can tell this is because there is not a VP that owns the entire product... It is a hodge-podge bunch of processes based on 1980's technology that anyone refuses to deviate from. The Sun guys have lost such control of their process and environment, that if someone wanted to change a single line in a source file for a driver, you have to recomple the entire operating system source tree (35,000 files) 3,400 makfiles to integrate your change. Almost all of their tools are written in bourne shell and ksh... Basically they have little chance of creating a high quality product and delivering that product to market at the right time at the right price.
Enough of HP/Sun... this is a systemic problem in the software industry and is getting worse. Configuration Management is an unknown discipline in Silicon Valley. Ultimately the software companies cannot manage change. I had the pleaure of working on a consulting project for a company that was trying to achieve CMM Level 2. They did... in record time... 14 months. At the same time they had been certified as CMM Level 2, they could not sucessfully compile their product during a 30 day period, install it or test it. Hows that for repeatability. Good marketing though
In summary, from the top of the food change CEO, to the bottom of the management food change (1st line managers), there does not exist the domain expertise to manage change. Not a clue! I know I've worked with these guys for 20 years!!! And quality is getting worse not better! Quality is not just bugs, its not being able to deliver what the customer requires.
The only revolution is going to come from the customer... not from the inside the industry
BP
Slashdot Mirror
I've been working in software organizations for over 20 years at companies like HP and Sun. Math and Science are not as important as good logic and tenacity! Most of the outsourced work to other countries is not complicated work... sustaining and testing is more about attitude than hard core math and science. That is what's mostly being outsourced.
Sun in the past has not been able to develop as desktop interface that gained any serious traction. Sun's offerings in the past have been Sunview, NeWs, OpenLook, CDE, GNOME. And now a Java 3d desktop... Sun has little experience dealing with end-users directly. At best I think they'll develop technology that some other entity (i.e IBM, BEA) will drive into the mainsteam. Sun dumped most of the 3D software engineers a few years back, loosing alot of domain expertise Not to sure the Linux world cares much about a Java Desktop, so Sun will cough up 99% of the engineering resources to build this stuff Seems to me new applications would be a significantly higher priority
Hey... you are not far off target with #4... The two most popular scripting languages at Sun used by the Solaris operating system organization is bourne shell and ksh... when I attempted to write to tools using perl I was told to use either sh or ksh... the computing environment at Sun in the solaris organization is based on 1980's technologies and tools... its really sad... basically they've built so much on-top of bad processes that everyone that wants to make a change, makes the change on top of the existing mess using outdated tools... And don't even get me going on the Java organization... its so confuscated people can't get their work done...
Last year I was on a project that required 17 Java programmers to complete a 3 month project that could have been completed by 1 PHP programmer and 1 systems programmer in 30 days... why would the company proceed in this manner. 1) its like a new tool-toy, its used for everything even when it should not be. 2) its politically correct. 3) its all part of the marketing campaign that must be one of the most successful in hi-tech...
Hmmm... so as I best recall from working at places like HP and Sun for most of the 80's and 90's... Unix Kernel developers were king of the hill. I'm talking about the folks that developed the VM system, Threading, Networking, core kernel work... (i.e not so much device drivers as lots on none core kernel folks contributed) Thru the late 80's into the mid 90's the graphics developers i.e hardware/software think framebuffers/opengl become very much in demand, required alot of math besides programming Core database developers have (i.e oracle kernel, sybase kernel) been hot the entire time In regards to OS Development vs IT careers there are alot of variables. They are such different disciplines and a very different culture. I've worked in both, started in IT (4yrs) then moved into engineering; database, networking, os, graphics (17 years). Finding both a culture and type of work can be challengine but very rewarding once you've found what meets your requirements. In regards to compensation, it really is supply and demand that drives the salary ranges. So in a long winded way... how to get started in OS development... At places like HP and Sun there are multiple tracks to doing this * Customer Support (in-house) people are assigned to specific areas of system functionality, i.e system commands, system libraries, networking, device drivers etc... * Sustaining Teams, they are supporting the customer support folks provide patches for released baselines * Technical Marketing, They are working closely with development and customers * Consulting Teams, They too are working closely with development and high profile customers * Any development team that is developming new hardware for an OS platform (graphics, storage, et c...) * An OS development organization... just get your foot in the door and the rest will happen as you develop your skills. You might consider (unit test development, system services development, buildmeister, lab administrator)
My experiences and observations based on working at Sun thru most of the 90's and recently on a contract is just the opposite. As people move up the corporate latter they become more conservative as there is more at risk, more to loose. Decision making takes longer and conflicts go unresolved because of the increased risk. A pattern of Good News up, Bad News down becomes the norm. And with the Peter Principle in full affect, accountability goes at the window. -psk
I wish I could... I don't know of any... Lots of great books have been written on Software Design and Methodologies. I'm not aware of any decent books on 'software infrastructure to support the software life cycle'. I'm inclined to think if anything of significance is going to happen in this area, it will be in open source. In regards to suggesting books... The classics by Pressman, Yourdan, DiMarco are good, so are the more recent books on xtreme-programming and also the books from MS (debuging the development cycle). But in terms of building a high performance software development organization... Nada -psk
I've been working in the computer industry since 1978, which included long stints at HP and Sun in the HP-UX and Solaris Networking/Operating Systems Organizations From my perspective the scenarios that are dicussed in this thread and many others are due to both the flexibility of software development and the rigidity required to develop the desired software, resulting in a continuous conflict that few teams are ever able to resolve. The processes and practices used by most companies in silicon valley are still stuck in the 1980's. The infratructures used to support development efforts at most companies are also stuck in the 1980's. Case in point: Most popular source control tools are based on SCCS, RCS and ClearCase (think Apollo DSEE from the 80's). Most popular editors are based on Emacs, Vi. Most popular IDEs are based on Microsoft and Borland from the 1980s. Most user interfaces are based on 1970's and 1980's look and feel (think HTML forms, think Microsoft, think Apple) One other key issues I've observed in silicon valley when it comes to software development: We have architects of technology but we don't have architects of product. To test this out in your situation, try to determine who owns the software development lifecycle and who owns the product. BigusKramicus
The person writing this article has the right idea in terms of software bugs, software complexity... but does appear to have a good understanding of what it takes to change the culture of a software organization let alone the software industry. In the commercial software business it starts with requirements management (which hardly exists) and Design. Design i.e architecture as a discipline ignores products and focuses on technology. Go to any software company in Silicon Valley and you will not be able to identify who owns product architecture, only who owns product technology. This results in a really interesting scenario where no one in the company really owns the product (i.e accountability). Since the product was never defined up-front and not owned by anyone, the quality is always going to be mediocre at best. If we take a quick stab at what the definition of product is: 1) application or system, 2) installation, upgrades), 3) administration, 4) documetation, 5) training, 6) customer support, 7) integration and consulting I've had the opportunity to work at HP and Sun in HPUX/SunOS/Solaris organizations and know first hand why quality is a problem... Its never been a priority. The way you can tell this is because there is not a VP that owns the entire product... It is a hodge-podge bunch of processes based on 1980's technology that anyone refuses to deviate from. The Sun guys have lost such control of their process and environment, that if someone wanted to change a single line in a source file for a driver, you have to recomple the entire operating system source tree (35,000 files) 3,400 makfiles to integrate your change. Almost all of their tools are written in bourne shell and ksh... Basically they have little chance of creating a high quality product and delivering that product to market at the right time at the right price. Enough of HP/Sun... this is a systemic problem in the software industry and is getting worse. Configuration Management is an unknown discipline in Silicon Valley. Ultimately the software companies cannot manage change. I had the pleaure of working on a consulting project for a company that was trying to achieve CMM Level 2. They did... in record time... 14 months. At the same time they had been certified as CMM Level 2, they could not sucessfully compile their product during a 30 day period, install it or test it. Hows that for repeatability. Good marketing though In summary, from the top of the food change CEO, to the bottom of the management food change (1st line managers), there does not exist the domain expertise to manage change. Not a clue! I know I've worked with these guys for 20 years!!! And quality is getting worse not better! Quality is not just bugs, its not being able to deliver what the customer requires. The only revolution is going to come from the customer... not from the inside the industry BP