Slashdot Mirror

It's not about what it can't do - it's about what it takes an inordinate amount of effort to do where it's a one-liner on every other modern operating system / scripting language known to humankind.

I really love instantiating .NET objects just to get a zip file from a web server, rather than curl http://server/path/to/file > file

It's worth noting also that he's not mentioned trying to get a response from Google (and if he's tried he's not included their response).

What response? Did you ever try to ask Google anything? First, any e-mail adresses of Google are surprisingly difficult to find... And second, if you do manage to find one, there will never be any response (even from those addresses that don't bounce).

If you've got a google address that responds, I'd be interested. Indeed, I'd like to discuss with them about their way of interpreting links into plain text (which short-circuits any rel="nofollow"), and also spams web server logs due to shorted http://server/.../page.html links.

And then, I'd like to ask how to make spam from google groups stop (apart from putting them into a Teergrube... but even that only cures the symptoms, as during the rare times where our Teergrube goes down, google groups spams still trickles through!)

Exactly. Hell, using a pair of cheap computers, two large capacity drives, and Linux, I can create a highly available NFS server, with seamless failover, so clients don't notice is the primary dies. Import that NFS share onto your webserver, map it to /downloads on your server, and voila! http://server/downloads gives you access to all the stuff stored on your nfs server, over the internet. Slap SSL into that equation, and nobody will be able to tell what you're doing on the server.

How's this one:
https://server/exchange/junkit/Inbox/Out%20of%20Office%20AutoReply:%20REMINDER:%2010_xF8FF_13_xF8FF_2007%209am%20EDT%20-%2010_xF8FF_14_xF8FF_2007%209pm%20EDT.EML?Cmd=open

They are referring to how OWA uses the Subject of a message as the URL. Notice the ':' in the standard Out of Office AutoReply message. It may not be an actual illegal character for a URL. But OWA does not make it easy to deal with. It would have been nicer if the MessageID had been used instead.

What does http://server/callhome do? I pointed my browser to it and I got httpd.apache.com 's site.

I have never heard of this. I get the Unix stuff for the crond.

someone steals your linux laptop. they are either so tech savvy that they know it would be stupid to leave the current OS in tact and format it, or they are computer *nix illiterate, can't figure out this damn linux thing, and install windows over your linux. likely the reason the os was left in tact was that it was a working windows install.

having said that your script accomplishes little more than the following command would also do if put in your cron.daily wget -O/dev/null http://server/callhome - i would recommend a dyndns domain in case you need to change the server your laptop is calling home to after your laptop has been stolen.

Huh, browsers default to port 80 for HTTP, so I think that you mean "http://server:port/whatever".

Don't use virtual hosts at all. If you can't get another machine, then at least run another copy of the web server on a different port. Then your production address would be something like http://server/whatever and the development would be http://serverport/whatever.

You are going to have to restart your web server in development more often than you would like to in production. You don't want to bounce production every time you change a module, right?

One easy way to accomplish this is to run cygwin http://www.cygwin.com/ on your windows box, setup all of your environment variables there (aliases and anything else you want), and have SSH pass your environment variables when connecting to a new box. With cygwin you get the standard openssh client, so anything you could do on a gnu box you can do there. You can get this all working and not have to change anything on the actual servers, it will all happen "automagically" on any server you log into, as long as sshd is setup to allow passing env. variables. Cygwin also lets you do all kinds of nifty things, like executing shell scripts from your windows box that log into the linux servers and execute commands. I've found that this is a great way to eliminate the human error factor when running identical commands on many servesr. Write a short script that will SSH in with SSH keys, run the script, verify the output/results and then display the results to the screen. Typically it doesn't save you too much time, as you need to write/test the script, but if can save you from making silly typos on production machines, and if your working on many servers, it can save time in the end. Nothing like a for do loop that iterates through all hosts and applies patches etc.

I also like one persons suggestion to put all the variables into a shell script on a webserver, and when you first log into a box you could just run wget -O - http://server/script.sh | bash or something along those lines.

Aliases in themselves are not a risk, it's only when they are impoperly used. For example, aliasing rm -rf / to something your might accidentally type isn't a good idea. However, if used carefully they can be very convinient and save quite a bit of time. I used them extensivly in my early unix admin days, but I found that they only caused me pain later when I would work on a box without my aliases and I'd be constantly retyping things and trying to remember what the actual command that I wanted was that my alias ran. Similar to people who don't know anyone's phone numbers because they are all stored in their cell's memory. It's great until you don't have it any you don't know any numbers.

Yeah, and those Location: headers are broken as well. Although most browsers accept them and act as implied, they really should specify fully-formed URLs -- i.e. beginning with http://server/ as opposed to a relative path fragment.

-b

Well I have used every macos version from 7 to osx, windows from 3 to xp, os/2, gnome, kde etc and my view is very different. I find that macosx and windows interfaces are almost unusable for me to get work done. I do db development for web apps and I spend a lot of time working with remote resources. In kde that is transparent. I can open/save with just about any protocol at any time. So I can just sftp to a remote server and just edit the file there and then hit save. It just works transparently and quickly. No latency from the network connection for editing so it works over slow or high latency links without any slowdowns that you see. I save several hours every week just from having url transparency and until you have REALLY used url transparency you just can't understand what it saves. I use sftp, webvdav, webdavs, ftp, http etc to work with files all the time. KDE even has those things so integrated in a file upload control for a web page you can put in ANY url that kde can understand and when you submit the page it will download the file from wherever that resource is and send it. Stuff like that I have not seen on anything else but I have heard that plan9 can do network transparency like that.

kde and gnome are even doing work to try and get their vfs layers merged and my hope is that eventually it gets moved to some kind of general filesystem layer so that linux will have everything as a url instead of just everything as a file.

Think of grep something sftp://server/path/to/file | webdavs://server:/path/to/result and have it just work. No need to lots of special tools to download and upload stuff.

Macs and windows look good but linux and kde allows me to get my work done in a lot less time and with fewer headaches. Some people seem upset with how customizable kde is because it makes it harder for people. From what I can see the defaults are set well but that customizability is stuff that I require. I use a 4 monitor setup that is tuned for exactly how I do my work and I hope to get more monitors at some point.