[Libre-soc-dev] SimdSignal library

lkcl luke.leighton at gmail.com
Tue Oct 5 01:06:06 BST 2021

On October 4, 2021 10:59:40 PM UTC, Jacob Lifshay <programmerjake at gmail.com> wrote:

>next: changing it to not use complex comprehensions.

hang on, jacob: remember i said, why are you spending time working on something that Cesar has already written?  that was a hint to stop working on it.  or you _can_ if tou like, but i won't - can't - authorise payment for duplicated work.

when i said (privately) that by working on a replacement that is perfectly functional and does the exact same job, not only can you not get paid for doing it (because we are paid for results, not by the hour) it also says, implicitly, that Cesar's work must be faulty, useless, or otherwise redundant.

you did not *explicitly* say this: it is *implied*, because i cannot think of any other reason left as to why time would be spent duplicating something that works perfectly and does the exact same job.

it was the same feeling for me when you went ahead and wrote the simd replacement library without discussion.

i felt, "what the hell, i spent 5 months on implementing this, and over a year on carefully designing it, what the hell is going on, he's not even looking at the code, the documentation, the bugreports.  is the code and the design *that* useless and worthless, he has to insult me by completely ignoring it?"

is there a specific reason why you feel it is necessary to do this?

it has been ten days, now, and you've not added one single line to the existing PartitionedSignal library, not helped at all with it.


the one valuable thing that you did though was to come up with the important insight on variable sized partitions.  this is extremely important, it will save an enormous amount of hassle, and solves an issue i couldn't handle.

however even with that insight, over the past 4+ days i've been the one that raised bugreports and documented it, putting down design thoughts and thinking through the caveats and practicalities.

i have some insights here as to why this is happening: it's a repeat pattern over 2 years so i think i have a handle on it now.

remember the Stage API? you *believed* i had not incorporated your input. actually what you did was write an exact parallel variant which was topologically identical, to the point where the demonstration examples that you wrote i was able to code-morph iteratively and quickly to the finalised API.

however the really strange thing is that you were *unable* to recognise that those examples were *your code*, would not work on them once they had been adapted, would not help add more examples, because you were *unable* to recognise that the underlying concepts were topologically directly equivalent.  that we were discussing the exact same design, just with different variable names and function names plus a few other minor alterations.

does that resonate at all?

what do you think is going on, here, and how can we sort it out?


More information about the Libre-soc-dev mailing list