Computing desk
< September 30	<< Sep \| October \| Nov >>	Current desk >

Welcome to the Wikipedia Computing Reference Desk Archives
The page you are currently viewing is a transcluded archive page. While you can leave answers for any questions shown below, please ask new questions on one of the current reference desk pages.

October 1

Part Number for Drive Tray used by HPE D2700 Disk Enclosure

I have bought a bare HPE D2700 Disk Enclosure (AJ941A), i.e. with only 2 SAS 6Gb Controllers, 2 PSUs and 25 Drive Blanks.
I intend to populate the D2700 with SSDs of my choice, but I first need to know the Part Number for the compatible Drive Tray, as the Drive Blanks cannot be used to install Drives. Vickreman.Chettiar 10:50, 1 October 2016 (UTC)[reply]

This forum thread may be useful. Tevildo (talk) 19:34, 2 October 2016 (UTC)[reply]

Is there a keystroke to return to 100% magnification ?

...using Windows 7, and typically Google Chrome. I am aware that CTRL 0 returns to "default zoom level", but that doesn't help. I want to be able to quickly go between 100% and 200% zoom levels, but CTRL + and CTRL - go in small steps: 100%, 110%, 125%, 150%, 175%, 200%. I currently have 200% set as the default, so that means I need many keystrokes to get back to 100%. I could set 100% as the default, but then I would have many steps to get to 200%. Ideally, 200% would remain the default, so CTRL 0 would still get me directly there, and I could get to 100% in one other keystroke.

Alternatively, if there's a way I can eliminate the intermediate steps, that would make me happy, too. BTW, if you're curious, my eyesight isn't so good, so I would prefer 200% all the time, except that an amazingly high proportion of web sites don't seem to handle that setting properly. Thanks, StuRat (talk) 00:38, 1 October 2016 (UTC)[reply]

See Windows key. ⊞ Win++ will zoom to 200%, ⊞ Win+- will zoom out if the Magnifier Utility is running, and ⊞ Win+Esc will exit the utility. Tevildo (talk) 08:15, 1 October 2016 (UTC)[reply]

Thanks, but I prefer to just use the zoom in the browser, to avoid the overhead from the magnifier utility, which also lacks the option to magnify the entire screen, under Windows 7, 32 bit edition. StuRat (talk) 16:20, 1 October 2016 (UTC)[reply]

OK. There is a Chrome extension called "Shortkeys" that appears to do what you want, but I've not used it myself and this isn't an endorsement. You can also zoom in and out with Ctrl-MouseWheel, which may be quicker than the keyboard, although it may not be obvious when you're at 100% rather than 110%. Tevildo (talk) 21:05, 1 October 2016 (UTC)[reply]

Thanks. CTRL mouse wheel works, and it lists the magnification level at all steps except the default zoom level, probably because that makes the "RESET TO DEFAULT" button (listed with the magnification) moot. StuRat (talk) 13:55, 2 October 2016 (UTC)[reply]

Incidentally, you might want to talk to your optician about getting a pair of bifocals, or a separate pair of reading glasses. I wouldn't be able to operate without them. Tevildo (talk) 08:22, 1 October 2016 (UTC)[reply]

You haven't read all my recent posts, I see. I have a skin condition which precludes me from wearing glasses, jewelry, watches, or even tight clothes. Therefore, I wear a single contact lens, so I have near vision in one eye and far vision in the other. (There supposedly are bifocal contact lens, but I am skeptical about how well they might work, so haven't tried them yet.) StuRat (talk) 16:17, 1 October 2016 (UTC) [reply]

Make building from scratch every time

I'm following the instructions here[1] to build Unity, specially using the "dpkg-buildpackage -rfakeroot -uc -b" command given. The first time building it took 25 minutes, which is to be expected, since it's a big program. No biggie. But my problem is that every subsequent build is also taking 25 minutes, meaning it's starting from scratch for some reason. I'm pretty sure it's building from scratch because I see messages like "[ 44%] Building CXX object launcher/CMakeFiles/switcher.dir/StandaloneSwitcher.cpp.o" when I haven't modified any cpp files at all, so no objects file should need to be recompiled.

How do I make it so that it builds incrementally? I.e. only compile and link the portion of the problem that that has been changed since the last build.

AFAIK make does incremental builds by default, and no sane person would turn off incremental builds without a very very pressing reason, so I'm at a loss as to why this is happening in the first place. Pizza Margherita (talk) 03:32, 1 October 2016 (UTC)[reply]

That's the default behaviour. To only do an incremental build, add the '-uc' option. From the man page

 -nc    Do not clean the source tree (implies -b if nothing else has been selected among -F, -g, -G, -B, -A or -S).

LongHairedFop (talk) 14:03, 1 October 2016 (UTC)[reply]

Thanks. Is it '-uc' or '-nc'? You said '-uc' but the man page says '-nc'. Pizza Margherita (talk) 16:16, 1 October 2016 (UTC)[reply]

With the '-nc' option it's not building from scratch anymore, which is great. But the bad news is that it's not building at all now. After a successful build with "dpkg-buildpackage -rfakeroot -nc -uc", I made some changes, and ran "dpkg-buildpackage -rfakeroot -nc -uc", but here's the result:

time dpkg-buildpackage -rfakeroot -nc -uc
dpkg-buildpackage: source package unity
dpkg-buildpackage: source version 7.4.0+16.04.20160906-0ubuntu1
dpkg-buildpackage: source distribution xenial
dpkg-buildpackage: source changed by Marco Trevisan (Treviño)
dpkg-buildpackage: host architecture amd64
 dpkg-source --before-build unity-7.4.0+16.04.20160906
 debian/rules build
dh build --with translations,quilt,python2,python3,migrations --parallel
 fakeroot debian/rules binary
dh binary --with translations,quilt,python2,python3,migrations --parallel
 dpkg-genchanges  >../unity_7.4.0+16.04.20160906-0ubuntu1_amd64.changes
dpkg-genchanges: including full source code in upload
 dpkg-source --after-build unity-7.4.0+16.04.20160906
dpkg-buildpackage: full upload (original source is included)

real	0m1.275s
user	0m1.096s
sys	0m0.068s

Basically the build system thinks nothing has changed, and that there's nothing to build. Pizza Margherita (talk) 21:32, 1 October 2016 (UTC)[reply]

php without script file

So I have a php script with;

<?php
echo hash('md5', $argv[1]);
?>

that gets called from the command line as so;

php.exe -f md5.php textgoeshere

which produces;

96ad4fad21a0d24c732a00cf3450e2ae

as the output. How can I do this with php directly on the command line bypassing the need for calling a separate .php script containing the function? Cohefuku (talk) 14:34, 1 October 2016 (UTC)[reply]

php -r "echo hash('md5', $argv[1]);" textgoeshere. -- zzuuzz ^(talk) 14:54, 1 October 2016 (UTC)[reply]

Wow, thanks zzuuzz! You're now officially a contributor to my new bot. I hope you'll be looking forward the to little surprise I have planned for the reference desk in the next few days... hehehe! Cohefuku (talk) 15:43, 1 October 2016 (UTC)[reply]

(edit conflict)This should work (works on Linux, I think it should be okay on Windows too):

   echo "<?php echo hash ('md5', 'textgoeshere') ?>" | php

-- Finlay McWalter··–·Talk 14:57, 1 October 2016 (UTC)[reply]

You can also do it a bit more simply on Linux systems like this

echo -n textgoeshere | md5sum

This will work on Windows too if md5sum is installed (although your version of course requires php to be installed). CodeTalker (talk) 21:01, 1 October 2016 (UTC)[reply]

Has any criminal been caught through machine learning?

I know that ML can be useful for detecting fraud or tax evaders. It could also be a good tool for concentrating police patrols there were more crimes could happen.

However, was any concrete case been solved by machine learning? --Hofhof (talk) 19:01, 1 October 2016 (UTC)[reply]

What an incredibly broad question! To help formulate your question better, you might want to look up some useful jargon and start thinking about the way our legal framework does things.

Cases are never "solved" - they are decided, or remedied, or settled. And evidence is never the solution, either!

Admissible evidence means some document, testimony, or tangible item that a court allows attorneys to present during a case.

Dispositional evidence [2] means that the evidence was in itself enough to set the result of the case.

Actionable intelligence means some fact or information - legal or otherwise - that lets a law enforcement agency make forward progress, possibly leading to a warrant or an arrest.

So, a better way to phrase your question might be:

"Has a machine learning algorithm ever yielded admissible evidence in a court case?" Surely yes, thousands of times; but it might take a while to find great examples. If we broadly consider that much of modern forensic evidence, particulary biometric data like fingerprints, results from computer systems that may incorporate machine learning methods, then it will be easy to find examples.
"Has a machine learning algorithm ever yielded dispositional evidence in a court case?" Surely yes, but it might take many hours reading boring court documents to track down good examples.
"Has a machine learning algorithm ever yielded actionable evidence leading to an arrest?" Surely yes, but the records might not be public unless a court releases them. Once again, it may take hours to track down great examples.

Many people speculate, for example, that Palantir Technologies employs machine learning engineers to create products that are used by the FBI to investigate white-collar crime and to investigate human-trafficking. Some speculate that this technology was also used to catch Usama bin Laden, a rumor often repeated by major news outlets like Forbes. I was not able to find any data that convinces me beyond a reasonable doubt - at least, not during a cursory search of the websites of the Department of Justice and the FBI's website. (I have looked before, and never found anything conclusive, either, but as they say, the absence of evidence is not the evidence of absence...)

Here is a case study from the FBI's website: Social Network Analysis: A Systematic Approach for Investigating. "With the support of robust technology, SNA becomes reliable across time, data, analysts, and networks and quickly produces actionable results inside any operational law enforcement environment." (The jargon in that sentence basically means that the FBI believes it can solve crimes using such algorithmic analysis). In this particular case study, the algorithmic information did not lead to an arrest or conviction, but researchers believe it could have led to arrest and conviction if administrative and legal policies were different.

Nimur (talk) 01:37, 2 October 2016 (UTC)[reply]

This isn't a field I understand that well, but it probably depends what you mean by machine learning. The controversial software TrueAllele uses Markov chain Monte Carlo and the Matlab Statistics and Machine Learning Toolbox [3] [4] [5] [6]. One of its competitors STRmix is programmed in Java and doesn't use Matlab but my impression is its design is similar [7] [8] [9]. Again I don't understand these that well but I don't know if you could say they were built using machine learning, and I'm fairly sure they don't learn between runs, but potentially you could say they use machine learning within a run. (The first external links for each software is the best explaination I found for how each work from a quick look.) Both BTW have been controvesially used in court cases in various jurisdictions including the US. Nil Einne (talk) 06:47, 2 October 2016 (UTC)[reply]

One may also want to be careful about what one means by ML leading to evidence. In many cases, it may be that the ML algorithms are used to uncover something that is subsequently reviewed by humans and the human interpretation rather than the computer one is what is presented in court. Circumstances where ML results are directly presented as evidence are probably quite a bit rarer than circumstances where they assist the development of a case in an intermediate fashion. For example, I don't know exactly how modern fingerprint database searches work, but I could certainly believe ML plays a role. However, when a fingerprint match is presented to the court, it is almost certainly based on the subsequent review and interpretation of a human expert rather than simply reporting what the computer generated. In part this is because of the role of cross-examination. Computers are complex, and some ML methods are quite opaque, which can make it easier for defense lawyers to cast doubt on their accuracy. So the prosecution is likely to rely heavily on human judgment when actually presenting a case before a judge and jury even if the initial observations leading to critical conclusions were first revealed by a machine. Dragons flight (talk) 07:15, 2 October 2016 (UTC)[reply]

Steve Jobs, from literature drop-off to tech entrepreneur

How could someone like Steve Jobs, with an educational background in literature (and not even a degree), become that successful in a technical field? --Llaanngg (talk) 19:17, 1 October 2016 (UTC)[reply]

Some would say he had a stronger background in electronics. There's a lot of people who have been successful at things which have nothing to do with their formal (institutional) education. -- zzuuzz ^(talk) 21:13, 1 October 2016 (UTC)[reply]

The technical stuff was done by other employees (originally Steve Wozniak). Jobs was just a manager cum charismatic guru. -- BenRG (talk) 21:28, 1 October 2016 (UTC)[reply]

That leads us to the question what he was doing there. Of the real work was done by Wozniak. — Preceding unsigned comment added by Llaanngg (talk • contribs) 21:53, 1 October 2016 (UTC)[reply]

He handled funding and manufacturing and marketing and sales. There's a lot to do in a tech startup other than laying out circuit boards and writing software. -- BenRG (talk) 22:07, 1 October 2016 (UTC)[reply]

For what it's worth, here's a source: a recent interview in which Woz said "Steve Jobs played no role at all in any of my designs" and "he did not know technology". I don't think this is a revelation to anyone in the industry: everyone knew that Woz was the engineer and Jobs was the businessman. -- BenRG (talk) 22:38, 1 October 2016 (UTC)[reply]

I think you are correct about people "in the industry", but anecdotally a lot of "average people" think Jobs invented the personal computer. Shows you how much of an impact image-crafting can have. This might be where the original poster is coming from. --47.138.165.200 (talk) 01:06, 2 October 2016 (UTC)[reply]

The original question asked how he became successful - and that question can be addressed entirely without reference to his technical skill. If you want to know about Steve's life, his authoritative biography, Steve Jobs, by Walter Isaacson, is worth reading.

Nimur (talk) 01:03, 2 October 2016 (UTC)[reply]

Also note that Bill Gates similarly is not so much a tech guy as a businessman, but recognized that the computer mouse (invented by others, such as Xerox) would be world changing, and made his software and hardware mouse-compatible starting in 1983-1984. StuRat (talk) 14:02, 2 October 2016 (UTC)[reply]

Bill Gates was more programmer than businessman in the early days. See Bill Gates#Early life. -- BenRG (talk) 03:10, 3 October 2016 (UTC)[reply]

(EC)Whoa back the truck up, by all accounts Bill Gates was a very skillful programmer. You can't put Bill and Jobs in the same basket. Bill became one of the most successful business men in history, but he started as "very much a tech guy", Jobs didn't even start as a tech guy. Vespine (talk) 03:11, 3 October 2016 (UTC)[reply]

It is also well documented by both Bill Gates and Warren Buffet that Bill tried to get Warren to be the owner and head of what would become Microsoft, but Warren didn't believe that he had the financing at the time. Steve Jobs, on the other hand, was already touting his concept of treating sales as a religion, encased beautifully in Klaatu's "Perpetual Motion Machine" lyrics. 209.149.113.4 (talk) 16:38, 3 October 2016 (UTC)[reply]

Because Jobs had the advantage of being both a great perfectionist for whom people would do their best work and also being around at the time when the state of the personal computer industry was so undeveloped that someone with minimal qualifications but considerable imagination could get into it and become successful in it. Similarly, Gates dropped out of Harvard slightly earlier because he felt that the developments of the period were so exciting that he had to drop out of Harvard and set up his own company before it was too late. Let's not forget that Jobs failed in many ways, too: when he set up NeXT a few years later, by the time Windows and MS-DOS had achieved dominance, it did not manage to win many customers - it was the Windows Phone 7 or WebOS of its day, admired but not used. It took Apple buying NeXT and making him CEO again to put its technologies to great use. There's an alternate universe where he gave up on it, Apple folded and he concentrated on running Pixar and that's what we remember him for. Blythwood (talk) 19:55, 3 October 2016 (UTC)[reply]