@MichaelHale Indeed. The electronic revolution happened so fast, that steely machines never had the chance to serve human with all their capacities.

hmm.. it cuts out the code - if u curious look it up n that post

@VitaliyKaurov I used the desktop version of Visual Thesaurus years ago, didn't know they're web-app now.

Interesting, I was unsure whether close votes did actually expire, here is more information

@JacobAkkerboom That information is a bit outdated. In addition to those points, if a question has been reviewed as "Leave Closed", then the existing close votes start getting aged even if the conditions in that answer are not satisfied.

@rm-rf thank you, that is also interesting and it makes sense.

Well, nevermind I guess, it seems that although I can paste the new contents of the clipboard back into MMA or textedit, I cannot paste them into chat or a Q/A, so it is broken anyway :)

Interesting Doc on Code Test, new in V10. See "Systematic Testing & Verification" part.

@Murta mostly old stuff except for EvaluationData and the Inactivate commands, or am I missing something?

*.wolfram.com SSL certificate has expired in the beginning of this week... just noticed.

Hmm, why is GraphPlot listed under code manipulation? Also VerificationTest etc seem new.

@JacobAkkerboom It works the same way as Test in MUnit (available with the Workbench). It looks like some of the MUnit functionality is rolled in to v10

@SjoerdC.deVries you are missing VerificationTest and TestReport. MUnit is coming in V10.

@kirma Something to do with zdnet.com/… ?

@rm-rf I don't think so. If the certificate was revoked some way, I'd expect it to show differently. Not to mention that wouldn't really remove the involved vulnerability.

Pretty appalling security hole nonetheless. The most interesting part about this is that coordinated vulnerability recovery effort clearly wasn't ready before the issue became public, and this puts more software systems at risk (in this case very big risk) than what would have happened if coordinated vendor effort would have been finished in time before full disclosure.

About this apparently enormous security hole called heartbleed, here is a thread from serverfault and one from reddit.com/r/programming and one from security

Digging my Mma on RPi. It'd appear it's not vulnerable... on grace of using OpenSSL 1.0.0d. Too old to have that vulnerability.

@kirma do you think that all rpi's have this old version, or just yours?

@JacobAkkerboom This is as new as is publicly available, as far as I know. OpenSSL (where ever it's needed, probably in HTTP functionality using libcurl in case of Mma) is rather conservatively updated library. Many vendors may be saved not being just pre-1.0.0, but actually running on versions like 0.9.8.

You can try something like find /opt/Wolfram -type f | xargs strings | grep 'OpenSSL [01][.]' on Linux, at least...

I meant pre-1.0.1, not pre-1.0.0. only 1.0.1 up to 1.0.1f is vulnerable.

@kirma thanks for the tip. Unfortunately my rpi is in another location right now, but fortunately it is no longer turned on working as a NAS.

@Murta Right. I overlooked those. Not sure if I will be using them. What about MUnit?

@belisarius, @bobthechemist, @rasher, @artes, @m_goldberg: I really feel this question should not have been closed. Could you reconsider your vote there?

@rm-rf See above, What do you think?

@SjoerdC.deVries Ok. I'll vote to reopen. But still, I can't repro here

@SjoerdC.deVries Can't repro here either, but reopened because you and Yves seem to be able to.

@SjoerdC.deVries Do you also get inconsistencies with Equal[]?

@belisarius @rm-rf The plot thickens. I tried all this code at work, where I have Mathematica running in a Citrix VDI (virtual PC), because of installation restrictions. The VPC runs Win7-32. I can produce the phenomenon easily there. At home now and on Win7-64. Nothing there. Looks like it's a Windows/Intel-32bit issue.

I'll dig up my old laptop, see what it does there.

@SjoerdC.deVries IMAP over SSL could be vulnerable with suitable version of OpenSSL on server, but I think Google never had vulnerable OpenSSL running on their services.

I already generate new passwords a few times per year for services I use. We might hear about something just as bad next week. So I change passwords on affected services and continue to try and improve and fix code I have control over. Should I feel OK about resuming thinking about more pleasant things?

@MichaelHale if you don't run a server, maybe

And if you don't have a bank account ?!?

@SjoerdC.deVries I really don't know.. I wish I did

turns out my mac has an old version of openSSL... but there are many things I don't understand

Luckily all my bank access goes through 2 factor authentication

So past communications could be decrypted. I don't know how that works. Are keys reused? If so, how long until they expire (and presumably deleted)?

@SjoerdC.deVries My bank uses a 2 factor auth too. They ask me for my mother's gender everytime I log on

@belisarius ;-)

I could make a very mean joke about that

@SjoerdC.deVries That was my very intention.

Well, the news has at least hit the front page of the bbc. Other than that, I don't see it much.

Bummer. The mma I wanted to test on my old laptop needs a new activation key. Don't dare to go to the Wolfram User Portal now.

@SjoerdC.deVries Ask baba ji for one. He's quick.

He's becoming a real regular

This is supposed to be a good check: filippo.io/Heartbleed/#wolfram.com it says wolfram.com is all good.

@SjoerdC.deVries I really wonder how many clients they get with such advs

Someone here with Python experience?

@JacobAkkerboom Is that site itself safe to visit?

@halirutan I'm going to learn it in a few weeks. Does that help?

@SjoerdC.deVries It would help in now + a few weeks :-)

@halirutan A colleague of mine is very enthousastic about it and giving a course

@halirutan My family in law are all reptiles and I deal with them frequently. Do you think my experience could help you?

@SjoerdC.deVries I have only very basic knowledge, but today I was helping some colleagues with volume processing algorithms. It turns out they need work with volumes of dimension 1500^3 and the only working knowledge they have is in Python. I thought that the speed issues cannot be that bad.. I mean people doing image processing with Java (imageJ) so it cannot be really worse in Python.

Unfortunately, I'm not able to get even a very simple loop over an array to an acceptable speed.

@halirutan Well, it is an interpreted language after all

There is a subset (pypy ?? ) that can be compiled

@SjoerdC.deVries I should note that Python needs 16 seconds, where I can get 3 lines Mathematica code doing it in 0.14 seconds

@SjoerdC.deVries Yep, was my idea too.. it was my second idea. My first one was to use numpy which basically uses C-arrays and provides a Python interface to them.

Unfortunately, pypy doesn't support numpy

Can you do their task in Mathematica and teach them as needed? Or will they not have enough kernels to do what they want?

@MichaelHale It's about a master-student of my colleague, so I think he should do it by himself. Since Python is the only thing he knows, I explained only the algorithms.

@SjoerdC.deVries I would think so. I saw it linked somewhere, but maybe that is not the best argument. Probably you are not using a bad version of openSSL yourself, but you should check. Indeed it seems that if you use a bad version of openSSL to connect, an evil site could read your memory. But I think the chance that you will use a bad version of openSSL is relatively small (if you are mac using chrome or firefox, it's probably fine). But I guess that site maybe only test if they run an old version..

and who knows if there is a man in the middle attack when you try to connect. I really don't know.

@halirutan Ah, I thought maybe they just needed a quick convolution or something. Learning Mathematica still might be worth it for the student in the long run though.

@belisarius @rm-rf Fired up my old laptop. The problem is easily replicated there. Conclusion for now is that it's a Win32 issue (or a a MMA32 issue)

@halirutan Hm... can you explain your point about Java and Python? I thought a typical for loop in python would be much slower in Python than in java, because Python is a dynamic language (like Mathematica I think).

@SjoerdC.deVries interesting

@JacobAkkerboom Yes, but with the numpy package you can force typed arrays and numpy has (a bit like listable functions in Mathematica) a vectorize method which takes a function and builds a listable version of it. Then you can write func(array) and func is applied to each element.

@SjoerdC.deVries I've had a similar problem on another platform years ago. It was a compiler bug, not cleaning up some registers before doing I-don't-remember-what

sometimes worked ok, sometimes don't. Depending on the initial state

@belisarius In the linked Mathgroup thread the Intel MKL libs seem to be culprit

Well, good night folks

@SjoerdC.deVries Good night

Good night!

I better go to sleep as well. Thank you for the explanation @halirutan, python is quite interesting to me.

@JacobAkkerboom Good night.