No. Size is not the enemy. Illegibility is the enemy. Every large code base that any of us have ever actually seen has been illegible, but that does not prove that it must be so. A code base of any size can be made legible by managing its complexity. This is one of the things that refactoring is supposed to do.
The motivation for seeking consensus is to minimize the number of people who might, at a later date, be tempted to throw oneself under the bus. This being so, it is a paramount organizational tactic and that fact will not change.
They have employees on the floor?? I've never seen one. If the story had been about new, improved ceiling cameras, I'd have believed it. But businesspeople's paranoia knows no bounds. This must trace back to one district manager being carpeted for a microscopic increase in shrinkage.
So you need to do a binary search but you don't want to use division instructions. The funny thing is that DEC's assembler for the PDP-10 did a very similar thing to set up the binary search of its symbol table, which was its critical path. The JFFO instruction (Jump on Find First One) was added to the PDP-10 instruction set for this purpose.
Binary search is easy because division by two is just a shift. It doesn’t need any division or even multiplication instructions. I can guess why they might have used JFFO but it seems needlessly complicated.
The trick is quickly finding the largest power of two that is less than the size of the table. It was done with two instructions: the second was the JFFO, the first was an FADD with a purposely denormalized operand. The symbol table was kept sorted on every insert.
That was my guess :-) But binary search doesn’t need explicit powers of two. I wonder if maybe the PDP10 is much faster at loops with explicit counters instead of comparing the search upper and lower bounds.
No, the loop would have been slower. That is why they found the starting point with that bizarre two-instruction sequence. For any table size between (2^^n)+1 and 2^^(n+1), inclusive, it returned 2^^n as the starting point. Then of course it had to loop after that.
Much mindwork can only be done in isolation. Academia today aspires to financialization, perhaps predictably by beginning with the playacting that results from the financialization of business and hoping that the money will follow the performance.
Extremely difficult task. Think of a kernel as a container of device drivers, where the motherboard/system is the top level device (scheduling, including interrupts), each processor is a device, memory is a device, etc. The most important decisions to make up front have to do with managing state across transitions between kernel mode and user mode. Say what you will of Windows, its model for this is potentially highly performant, though Microsoft's documentation is so poor that most drivers are badly written: too much is done in ISRs versus DPCs. (In the early days, Microsoft sold very expensive training under NDA; I'm not sure they are still doing that but it is why, to this day, the doco still leaves out so much essential context.)
Red tape actually implies excessive standards that probably doesn't lead to any real accountability gains. You can have a lot of red tape with all bad rules, for example.
I know when it is used as a slogan at political rallies, and they just simply call out things like "these regulations, so stupid, why can't we dump waste in the river, they are reducing our profits".
It is all about profit. The environment, or safety, anything in the way of profit is called '"red-tape" and is put there by some do-gooder.
So, yes, I think I can infer when someone talks about getting rid of 'red-tape' what they are "meaning". But that is also why I listed several categories of red-tape, since some 'red-tape' should be updated.
The First Law of Modern Politics is that nothing may ever be called by its right name. This applies recursively, i.e. it applies to metaphors and euphemisms as well as to their original referents. "Red tape" means "you are making it harder for me to steal money, cut that shít out."
reply