* [dts] Runtime code generation for RTE Flow
@ 2020-10-07 20:13 Owen Hilyard
2020-10-09 6:16 ` Ma, LihongX
0 siblings, 1 reply; 2+ messages in thread
From: Owen Hilyard @ 2020-10-07 20:13 UTC (permalink / raw)
To: dts, Tu, Lijuan, Ma, LihongX, Lincoln Lavoie, Sarah Hall
[-- Attachment #1: Type: text/plain, Size: 1506 bytes --]
Hello all,
I've been working on tests for the flow API recently, and I've hit a point
where I think runtime code generation would be useful. For context, the way
the test suite is architected right now, I start up testpmd in the
set_up_all function, and then every single flow rule that is tested get
it's own test case. I did this so that we can have the benefits of only
starting testpmd once, but we can still have the granularity of knowing
exactly what failed. Currently, I have a script which generates all 54
pattern matching test cases and prints them out, this means that I can copy
and paste the generated test cases into the test suite. I am concerned
about someone who needs to maintain this test suite afterward not receiving
that important bit of information. The simple solution to this problem that
I see is to modify the test suite to add the test cases at runtime.
The reason I'm reaching out instead of just doing this and submitting this
is that I wanted to make sure that there no strong objections before
starting. Doing this would probably involve the use of exec to avoid
needing to hook parts of the interpreter to generate the symbol tree
directly. Performance isn't an issue since this takes roughly 3/1000 of a
second.
More details on the plan for the test suite can be found at
https://docs.google.com/document/d/1_jEciQFZ-Lj1ASF_mQbCnB3U5FefdW2Z84HP1y-GJek/edit?usp=sharing
I'd appreciate any concerns or suggestions on this.
Owen Hilyard
UNH InterOperability Laboratory
[-- Attachment #2: Type: text/html, Size: 1717 bytes --]
^ permalink raw reply [flat|nested] 2+ messages in thread
* Re: [dts] Runtime code generation for RTE Flow
2020-10-07 20:13 [dts] Runtime code generation for RTE Flow Owen Hilyard
@ 2020-10-09 6:16 ` Ma, LihongX
0 siblings, 0 replies; 2+ messages in thread
From: Ma, LihongX @ 2020-10-09 6:16 UTC (permalink / raw)
To: Chen, Zhaoyan, Peng, Yuan
Cc: Owen Hilyard, dts, Tu, Lijuan, Lincoln Lavoie, Sarah Hall
[-- Attachment #1: Type: text/plain, Size: 1884 bytes --]
+ zhaoyan and yuan
How do you think of this?
Regards,
Ma,lihong
From: Owen Hilyard <ohilyard@iol.unh.edu>
Sent: Thursday, October 8, 2020 4:14 AM
To: dts@dpdk.org; Tu, Lijuan <lijuan.tu@intel.com>; Ma, LihongX <lihongx.ma@intel.com>; Lincoln Lavoie <lylavoie@iol.unh.edu>; Sarah Hall <shall@iol.unh.edu>
Subject: Runtime code generation for RTE Flow
Hello all,
I've been working on tests for the flow API recently, and I've hit a point where I think runtime code generation would be useful. For context, the way the test suite is architected right now, I start up testpmd in the set_up_all function, and then every single flow rule that is tested get it's own test case. I did this so that we can have the benefits of only starting testpmd once, but we can still have the granularity of knowing exactly what failed. Currently, I have a script which generates all 54 pattern matching test cases and prints them out, this means that I can copy and paste the generated test cases into the test suite. I am concerned about someone who needs to maintain this test suite afterward not receiving that important bit of information. The simple solution to this problem that I see is to modify the test suite to add the test cases at runtime.
The reason I'm reaching out instead of just doing this and submitting this is that I wanted to make sure that there no strong objections before starting. Doing this would probably involve the use of exec to avoid needing to hook parts of the interpreter to generate the symbol tree directly. Performance isn't an issue since this takes roughly 3/1000 of a second.
More details on the plan for the test suite can be found at https://docs.google.com/document/d/1_jEciQFZ-Lj1ASF_mQbCnB3U5FefdW2Z84HP1y-GJek/edit?usp=sharing
I'd appreciate any concerns or suggestions on this.
Owen Hilyard
UNH InterOperability Laboratory
[-- Attachment #2: Type: text/html, Size: 4735 bytes --]
^ permalink raw reply [flat|nested] 2+ messages in thread
end of thread, other threads:[~2020-10-09 6:16 UTC | newest]
Thread overview: 2+ messages (download: mbox.gz / follow: Atom feed)
-- links below jump to the message on this page --
2020-10-07 20:13 [dts] Runtime code generation for RTE Flow Owen Hilyard
2020-10-09 6:16 ` Ma, LihongX
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).