Club409
Calling on you computer nerds
every time you do that the terrorists win
that's how i do it too unless it's a really short string then i just paste it into the "Run..." dialog and recopy :)
You can paste into a plain text editor like notepad but I suspect that there's a more elegant solution.
For short text copied from a browser I just paste into the url or search field. CTRL+C text CTRL+E CTRL+V CTRL+A CTRL+C then CTRL+V to destination. :O CEVACV. It's how I fund my local tucker YMTA facility. Young Muslim Terrorist need a safe environment to hone their swimming racquet ball and jihadist skills too. :O
Also with office 2010 you can right click and paste plain text. Subsequent pastes in the same document remember that option. You can probably set a preference to always paste plain text.
I came up with a pretty slick bug workaround today. One of my coworkers was attempting to run a calculation on about 120000 processors and it kept dying with an error message from some 3rd party software that we make use of. There was a check of the form (length_of_small_vector < length_of_big_vector) that was evaluating to false when the small number was about 15 and the large number was around 3.2 billion. We realized the issue was that they were using a standard int to store the sizes and thus 3.2 billion was overflowing to a negative number. If we had access to the source of that package we could just change int to long and be on our way but that's not an option and getting them to change their code and getting the new version installed could take weeks. Also decreasing the size of the problem so that length_of_big_vector < 2^31 (about 2.2 billion) would invalidate the results. I half jokingly suggested he just increase the problem size to that the vector length was more than 4.4 billion and it would overflow so far that it would be positive again. There was a silence for a few seconds followed by "that just might work" and a resubmitted job a few minutes later.
nice you were always the smart one steven
So they could store a vector with more than 2^31 entries but not report its size correctly? that's an odd situation. way to go macguyver.
Would it save more memory to increase the large vector or just increase the size of the small vector to the point that it is also negative but more negative than the large? </optimizing sad situation>
the creation of vectors is all done locally (i.e. on each processor) so as long as the length of the vector per processor is under 2^31 then there isn't a problem. they just happened to have a value that represents the cumulative size of all of these local vectors that isn't functionally necessary and they performed some sanity checks which end up failing because the total length was too big.
the small number isn't actually a vector (i wrote that incorrectly) it's the number of big vectors that you have to store. increasing that number increases the memory requirements.