Follow Slashdot blog updates by subscribing to our blog RSS feed


Forgot your password?
Note: You can take 10% off all Slashdot Deals with coupon code "slashdot10off." ×

Comment Re:HTTPS everywhere (Score 1) 174

Set up transparent proxy, redirect all connections to outside server ports 443 to the transparent proxy server. Set the proxy (squid can do this) so it generates a certificate on connection using your own self-signed CA certificate to sign it. Install the CA certificate on your PC.

Now your proxy server is the man-in-the-middle - having encrypted connection from it to you and from it to the origin server, but also having access to the unencrypted content. This is exactly like a MITM attack, except you are doing it on yourself.

I have set this up for one client - in that case the proxy is there to 1)log what sites are visited and 2)block some sites (like facebook and youtube). Doing it with the proxy is better than just having a blocked IP list, because those IPs change and sites like youtube or facebook have lots of ips. I can not parse the access log, find what other sites I should block and block them.

Comment Re:Where is Commander Adama when we need him? (Score 1) 189

No, the automatic choke in my car is just a bimetal spring that, when cold, opens a valve in the carburetor allowing extra fuel to be drawn into the intake manifold. The spring is heated by the antifreeze from the engine and an electrical heating element.

No netwoek necessary. No computer necessary. Or any electronic components except a piece of nichrome wire as a heating element.

Comment Re:TIP series are good devices (Score 1) 170

I have a desoldering gun, it does not take a long time to desolder a big DIP chip. While most parts are cheap, I still get the inconvenience of having to buy one, so I'd rather use sockets.

I have a hot air gun, but so far, my rate of success with it (managing to desolder a chip without destroying a nearby chip) is not that great.

Comment Re:TIP series are good devices (Score 1) 170

I guess it is easier to me to solder bigger components. If I make a mistake (or want to improve the device) and need to desolder a part, with SMD it is even harder to do properly (without destroying the part) than soldering it. Something like SOT is more difficult than TO92. and DIP is easier than SOIC and much easier than TQFP, especially desoldering. Also, with DIP I can use sockets (that are cheap). PLCC would be OK too, but not many modern components are PLCC.

And if I do something with tubes I do not even need a board.

Comment Re:TIP series are good devices (Score 1) 170

The TO-3 part was a response about size. And while I am able to etch a single sided board, I found it is too much hassle for me compared to using a perfboard since I am not making multiple copies of the same device.

And if I wanted to build a transistor amplifier (I recently completed a tube amp that does not use any semiconductors, and plan to build a couple more (but different)), I would most likely use linear design, not switching. Disregarding any comments about sound quality (which depends on more things than just the topology of the amplifier), the switchers produce a lot of EMI - I have bought a small class D amp on a board and while it sounds OK, it produces so much EMI that if I keep it near my KVM switch, the image on my monitor (VGA connection) becomes distorted and when the amp is on my FM radio picks up more noise. Of course, that can be solved by shielding the speaker cables and the amp, the linear amp does not require as much shielding. The switchers also clip even worse than class B transistor amps.

The only thing good about class D amps is that they use less power. However, if I am not powering the device from a battery and/or do not intend to keep in on 24/7, I do not really care about the power.

Then again, I have a tube headphone amp that I use a lot and keep 24/7 unless I plan to not be at home for several days.

Comment Re:old clunky junk (Score 1) 170

It is still more difficult and also the chip needs to be programmed in the first place. The programmer also may not be cheap. So, if all I need is to flash a LED, I'll use two transistors or a 555 chip. For something complicated, yes, a MCU is better than a board full of 74xx or 4xxx logic. And for stuff that needs Ethernet connection, I'll just use Raspberry Pi.

Comment Re:old clunky junk (Score 2) 170

Programming microcontrollers is difficult, especially debugging the program (it's not like with a PC program where you can use a debugger or log everything etc), Arduino makes it a bit easier. Discrete logic is easier still, but if you want to do anything complicated the 74xx chips take up a lot of space.

Also, there is such thing as "good enough". If I want to drive a nixie tube (or something else that requires relatively high voltage and low current), I'll use a MPSA42. Why? Because it's cheap and works OK for that purpose. I am not buying millions of them so even if some other part is $0.01 cheaper, my time wasted evaluating and selecting for it is worth more than the $0.1 I'll save buying 10 of them. Quite a few times I buy what the local store has in stock instead of special ordering a part. If the in-stock part can work where I need it to, then OK.

Comment Re:TIP series are good devices (Score 2) 170

And yet, carburetors are used for a lot of engines - small engines such as those in lawn mowers or chainsaws and also in light aircraft engines.

Sometimes the smaller device does not mean better - SMT parts, for example, are more difficult to solder pretty much require etching a PCB (instead of using a perfboard or point-to-pont construction). Also, if I am building an electronic load or an analog amplifier, the transistors are going to dissipate a lot of heat anyway, so I might as well get a TO-3 part.

Comment Re:Now that's just evil (Score 1) 515

"Hey! Your NES is total crap, uses lots of power, has a lot of bugs, get a PlayStation, it works better."
"Will I be able to play Mario on it?"
"Well, I really want to play Mario, so your new system is totally useless to me".

Same with PC OSs - replacing Windows with Linux will only work for a user, if all the software the user needs either runs on Linux or has good (meaning very similar to Windows version) Linux alternatives. You can make Linux UI (KDE, LXDE) look pretty much like Windows (up to 7), which is good. But if the user needs, say, AutoCAD, and AutoCAD does not run on Linux, then even though Linux is "better", "free" and "more secure" it still is useless to the user.

Comment Re:Who is evil? (Score 1) 515

You see, in theory it may be like this, but in practice it's different.

There are standards, and even if there is not actual standard, there usually is a de-facto standard. Windows desktops are one such standard. Because of good backward compatibility (not perfect but certainly better than Linux or MacOS), Windows software usually runs on a few Windows versions, even if those versions have age difference more than 10 years (quite a lot of software runs on XP and on 10). This creates an expectation. Linux completely has the right to not be compatible with any Windows software (after all, Windows is not compatible with Linux software). However, that creates a difficulty in switching from Windows to Linux. And usually the user does not care whose fault really it is, and assigns the blame to the "new" (to him) option.

An example: MS Office does not run on Linux, but Libre/Open Office is almost as good. However, it sometimes has problems with MS Office file formats. While the blame should go to Microsoft for having poorly documented formats, I, as a user, cannot make MS change the formats or make LibreOffice better support the formats that exist. However, if my clients use MS Office (and its formats), I have to be able to open the files they send me, which means I have to have MS Office.

After all, what can I do to make MS change the formats or whatever? Call Putin and ask him to threaten MS with a nuke unless they change? It's not like Putin would even talk to me.

Comment Re:Now that's just evil (Score 2) 515

One of our clients has dovecot now. They use Outlook on the client side and have multiple mailboxes shared between multiple users (user A may use mailboxes I,J,K, user B may use mailboxes K,L,M etc).. Most of those maiboxes contain gigabytes of emails. Outlook downloads the mailbox contents to the server (they use Terminal services - all users are on the same server). Which means that now there are multiple copies of the same mailbox on the server really wasting space (200-300GB or so) not to mention the copy on the IMAP server. Then if you, say, move a large amount of emails from one mailbox to another, it can be really slow with Outlook essentially freezing for a long time. Oh, and if one user quits the job, you need to change passwords on the mailboxes that the user had access to and then configure the new passwords for essentially everybody else.

With Exchange, the users each has his own password and can be assigned "rights" to some other mailbox. The email is kept in the Exchange server, no more hundreds of gigabytes of duplicate files.

"Here's something to think about: How come you never see a headline like `Psychic Wins Lottery.'" -- Comedian Jay Leno