http://loader.gadgetzone.com.au/Movies/July-2008/20-things-that-Windows-7-MUST-include.aspx?Page=1
Gadget Zone has a wonderful writeup on what sorts of things Windows 7 truly ought to include. He hits some wonderful high points, and the article is well written, which is truly refreshing in the world of modern hardware reviews. These days, it seems to take six pages to say “Card X is basically faster, but only half the price of Card Y.”
So in the spirit of a true Bertrand Russell wannabe, I’m going to comment.
After all, not only is there an elephant in the room when it comes to features Windows 7 should have, there are several. A treatise on minor annoyances like standards-compliant web browsing brings the Titanic’s deck chairs to mind. I’d like to volunteer a few suggestions for features for the upcoming Windows 7 in a spirit of true sarcasm and disgust. Not for you, Mr. Casissi, but for the hypothetical operating system that shall one day be foisted upon us all.
Yes, foisted upon us all — http://www.networkworld.com/community/node/29644?ts
Reasonable Hardware Requirements. Windows 7 should be available to function smoothly and comfortably within the confines of a gigabyte or less of RAM, a 32-bit processor, and a hard drive small enough that it can’t contain every work of literature ever produced by the human race. Windows Vista finds this challenging, but these are the sorts of things that will quickly alienate Linux users everywhere.
Lack of Virus Support. Linux has exploits. There was one discovered recently surrounding DNS. Linux is vulnerable to DOS attacks. What Linux does not include is backward compatibility for every virus ever invented. It’s been a couple of weeks since the DNS exploit has cropped up, and the Linux developer community is on it like a pack of rabid dogs smelling blood followed by a gaggle of angry Klingons sensing a slight against their honor. Someone just unveiled a processor-level exploit in modern x86 hardware that doesn’t care about the OS. And I’m certain Linux will be the first platform with a fix. Which segues nicely into my next suggestion.
Open-source TCP/IP Stack and NIC Drivers. Open-source Javascripting. Why are Linux developers around the world feverishly chasing an obscure exploit? Purity and security. The exploit is a rare breed, and they hope to keep it extinct. With the huge preponderance of attacks being internet-based, a simple switch to open source would cut off the primary vector of infection and Windows Security issues. There is a fleet of Windows developers trolling about on MSDN who would no doubt love to contribute to the Windows Security effort. If that were possible.
Ground Level Rewrite. Windows is so crufty these days, it’s insane. Is it really necessary for the average user to have tape drive drivers installed on their hard drive? Does a typical computer these days have a pressing need for ISA support, or a PC speaker device listing? Or mountains of C? Microsoft should, by necessity, be leading the race in developer toolkits and IDE’s, just to manage their huge mountains of code. You could hide the entire Bible three times over without being noticed. The only reason they’re not leading the pack for making life easy on coders is their overwhelming chore of managing what they’ve already got. It’s exorbitant at this point to do something truly innovative with Windows, like trying to make a u-turn in a train.
Global Driver Repository. A fresh installation of Windows already reboots something like six times on average before actually getting to a desktop screen, whereafter the user may begin installing their own useful software and rebooting volutarily another six or thirty times. So what does Windows actually need to have available? Basic storage drivers and a mind-blowing list of NIC drivers. It needs to be able to interact on some level, even a basic one, with the target drive. Thereafter, with an internet connection and a global driver repository, it can use hardware ID’s to upgrade to its heart’s content. Linux is already evolving in this direction, because it’s the next logical step.
Faster, Smaller Software. I downloaded OpenOffice the other day, and it was a little over 100mb. Office doesn’t even fit on a single CD anymore. Irfanview has yet to top a half megabyte, in the age of terabyte hard drives and gigabit ethernet and petaflop supercomputers, and easily outperforms every image utility released with Windows since its inception, in terms of size, usability, features, and speed. And it’s free. Firefox is up to 17.2M for its Macintosh version, the largest of all. The download for Internet Explorer 7 is 14.2M on its home platform, Windows on x86. But don’t be fooled. Even though it’s only 2M bigger, it’s because that’s to install the installer. Microsoft won’t release a redistributable installation package. The last full redistributable was version 6.0 at 76.7M.
User Mode. For something like a Windows XP virtual machine or DOS-mode software support, libraries need to be called, and a compatibility layer enabled. Temporarily. For that program. And highly optimized. The notion of leaving DOS support available for the operating system, or a Windows XP virtual machine, is ridiculous, like installing a kernel module in Linux to enable direct execution of Win32 binaries. Any backwards compatibility, emulation, or cross-platform support should be strictly in user mode for the sake of security and stability. Want to know Windows’ dirty little secret? To date, the entire system slows to a crawl to accommodate kernel-mode network and sound drivers. Try running applications on a fresh install before NIC and sound are installed. It’s fast, like driving a stick of butter across hot teflon. Run these drivers in user mode, and aside from a huge security boost, you get your system back ala country music.
Scriptable Window Manager. This one thing alone costs me 12 minutes a day at work. And that’s after I’ve scripted as much as I possibly can. I have to cram four roughly 800×600 windows into one desktop such that all are mostly visible at one time, and each morning an entire series of windows have to be resized and moved.
Stability. Many complain about the amount of time required to go from POST (Power On Self Test) to a usable Windows Desktop. But stability would eliminate the need for a speedy boot or reboot, as would limiting the number of processes and files that could precipitate a system restart. Most users wouldn’t mind a full ten minute startup if it were only necessary once every six months.
The current state of the art isn’t terribly artistic. When words like “bloated” and “monolithic” are thrown around as buzzwords, the classic corporate pitfall of huge failures mitigated by mountains of liquidity become obvious. Microsoft once stumbled upon a goldmine, but they’ve elected to not evolve and create a leaner meaner product, choosing instead to grapple with an enormous body of work by way of minor tweaks, patches, fixes, and upgrades.
Windows 7 needs, more than anything, a fundamental change of philosophy.