Date: Sun, 09 Apr 2023 17:43:13 -0300
On Sunday, 9 April 2023 15:57:38 -03 Phil Bouchard wrote:
> Well a demo license is as free as anything Open Source.
Free as in beer. Not free as in speech and I need the latter. If you impose
any conditions on what I may do with the information I learn by looking at it,
then I can't look at it. It's simple bureaucracy: unless it's one of the OSI-
approved licences, I can't accept it on behalf of my employer.
> It is thread safe because it is thread-local. So it depends on the use
> case but if you only need threads-local FIFO containers then you can
> easily make use of it.
The standard Linux malloc() is also thread-local (for the most part) during
allocation. If you're getting a 3.6x performance over it, then one of the
following must be true:
a) you've come across an extremely important improvement that others haven't
seen
b) you've hyper-optimised for one or a few use-cases, to the detriment of the
majority
c) you've missed something really important that has caused the others to
require a much larger overhead.
If it is (a), it will require a full paper explaining just what that is (with
whichever protections you may want for your IP). If it is (b), it deserves a
paper explaining when it should be used and when it shouldn't. And if it is
(c), you need to know about it and fix it.
> Well a demo license is as free as anything Open Source.
Free as in beer. Not free as in speech and I need the latter. If you impose
any conditions on what I may do with the information I learn by looking at it,
then I can't look at it. It's simple bureaucracy: unless it's one of the OSI-
approved licences, I can't accept it on behalf of my employer.
> It is thread safe because it is thread-local. So it depends on the use
> case but if you only need threads-local FIFO containers then you can
> easily make use of it.
The standard Linux malloc() is also thread-local (for the most part) during
allocation. If you're getting a 3.6x performance over it, then one of the
following must be true:
a) you've come across an extremely important improvement that others haven't
seen
b) you've hyper-optimised for one or a few use-cases, to the detriment of the
majority
c) you've missed something really important that has caused the others to
require a much larger overhead.
If it is (a), it will require a full paper explaining just what that is (with
whichever protections you may want for your IP). If it is (b), it deserves a
paper explaining when it should be used and when it shouldn't. And if it is
(c), you need to know about it and fix it.
-- Thiago Macieira - thiago (AT) macieira.info - thiago (AT) kde.org Software Architect - Intel DCAI Cloud Engineering
Received on 2023-04-09 20:43:19