Hi Enrico,
Please restrain yourself from sharing your opinions quite so strongly in
the future. This mailing list is shared by project enthusiasts like
yourself, core team members and contributors, and people who are often
making their very first contribution to an open source project. This can
be a daunting prospect, and it is a shame to undermine the experience by
having their first contact with the project be someone informing them
that they're going to be doing it wrong.
As to your technical arguments, I know that you are strongly in favour
of loosely-coupled components. I believe the design described in the
wiki page plays to the strengths of Rust's type system, and I think the
result will be more maintainable and than a logging-based one, as well
as allow better introspection at runtime.
Cheers,
Josh
On 3/10/18 12:37 AM, Enrico Weigelt, metux IT consult wrote:
On 09.03.2018 19:52, Patrick Poggi wrote:
Hello,
Bhanu and I are both NCSU student that will be working on the http
archive
support project.(
https://github.com/servo/servo/wiki/HTTP-archive-support-project)
At the moment, we are unsure of how to proceed with the project. Any
suggestions or direction would be helpful, as this is our first time
contributing to an open-source project.
Isn't that just writing detailed log files ?
I'd seriously suggest *NOT* writing HAR files directly, *NOR* collecting
all records in some central buffer - introduces all sorts of concurrency
problems (not even speaking of multiprocess issues) as well as inter-
dependencie of many subsystems, which in turn requires much lock/wait
issues, therefore distorts the measurement itself.
Instead just add simplicistic logging to various areas you're interested
in (please make it compile-time configurable - at least until we have an
kprobes-alike runtime code-path configuration ...) and take care that
the log-writing is as fast and simple and non-intrusive as possible.
Just write out the few bits (w/ exact timestamps), standard formats, etc
(could also be arch-specific binary encoding). Pulling things together
into formats like har, human-readable logs, etc, should be done
separately - that code should be available as a separate command line
program (one might wanna run this on separate machines).
The approach described in the Wiki page is unnecessarily complex and
risks adding a lot latency and subtle bugs. I'm opposed to that.
--mtx
_______________________________________________
dev-servo mailing list
dev-servo@lists.mozilla.org
https://lists.mozilla.org/listinfo/dev-servo
