3 ms·

"On a small scale, a classic yields significantly different meanings when read in different circumstances and moods; on a larger scale, a classic conveys wholly

by thret 15y ago

"On a small scale, a classic yields significantly different meanings when read in different circumstances and moods; on a larger scale, a classic conveys wholly different worlds when read in different times of life, at different stages of experience, feeling, and understanding of life. Classics may be interesting and even entertaining, but people always find they are not like books used for diversion, which give up all of their content at once - the classics seem to grow wiser as we grow wiser, more useful the more we use them." - Thomas Cleary

Monkeyget 15y ago
I read MMM years ago when I was in school because it was supposed to be one of the must read book of the field I was getting into. I found it abstruse, outdated, irrelevant and boring at the time. The older, more mature, me with some real world experience, got back to a completely different book. What was dry is now engaging. Random opinions are now insights. It also helps that my English is much better. The obscure vocabulary and outdated terms have become good writing and old fashioned (in a good way) style. One thing that surprises me is the definition of architecture. To Brooks the architecture is the set of manuals needed for the user to use the system being created. It sort of is the (external) specification of the software. He clearly states that the architects must refrain from prescribing how the implementation should be. This is very different from today's definition of architect and software architecture. Why is it defined like this? Why isn't the conception of the (internal) architecture of the software a concern to Brooks?
- jmount 15y ago
  I missed the change in meaning of architecture until you pointed it out. I had an uneasy feeling about the use of the word when reading, but you have really clarified what the difference in use is. Thanks!
- rg 15y ago
  Brooks particularly had in mind IBM's System 360. One of its goals was to have a single architecture span an extremely wide range of performance. So there were many models (e.g., 360-20, 360-65, 360-90) with a couple of orders of magnitude differences in cost and performance, but sharing a single architecture. Implementation teams built each model with attention to its specific required performance and cost constraints, leading to many different hardware designs. The system architects worked to define an architecture without constraining the possible implementations unnecessarily.
- Someone 15y ago
  Yes, but it is what the 'A' in ISA stands for (http://en.wikipedia.org/wiki/Instruction_set_architecture http://en.wikipedia.org/wiki/Instruction_set_architecture)

by thret 15y ago

Monkeyget 15y ago

I read MMM years ago when I was in school because it was supposed to be one of the must read book of the field I was getting into. I found it abstruse, outdated, irrelevant and boring at the time. The older, more mature, me with some real world experience, got back to a completely different book. What was dry is now engaging. Random opinions are now insights. It also helps that my English is much better. The obscure vocabulary and outdated terms have become good writing and old fashioned (in a good way) style. One thing that surprises me is the definition of architecture. To Brooks the architecture is the set of manuals needed for the user to use the system being created. It sort of is the (external) specification of the software. He clearly states that the architects must refrain from prescribing how the implementation should be. This is very different from today's definition of architect and software architecture. Why is it defined like this? Why isn't the conception of the (internal) architecture of the software a concern to Brooks?

jmount 15y ago
I missed the change in meaning of architecture until you pointed it out. I had an uneasy feeling about the use of the word when reading, but you have really clarified what the difference in use is. Thanks!
rg 15y ago
Brooks particularly had in mind IBM's System 360. One of its goals was to have a single architecture span an extremely wide range of performance. So there were many models (e.g., 360-20, 360-65, 360-90) with a couple of orders of magnitude differences in cost and performance, but sharing a single architecture. Implementation teams built each model with attention to its specific required performance and cost constraints, leading to many different hardware designs. The system architects worked to define an architecture without constraining the possible implementations unnecessarily.
Someone 15y ago
Yes, but it is what the 'A' in ISA stands for (http://en.wikipedia.org/wiki/Instruction_set_architecture http://en.wikipedia.org/wiki/Instruction_set_architecture)