Re: Mynewt logo - request for input

[Christopher Collins]

I think they are all really good, but my preference is for 1 or 2.

I lke 1's font the best, and I like the way the newt's tail wraps around
the y.  So actually I guess I like 1 the best :).

Chris



On Tue, Nov 17, 2015 at 09:49:24AM -0800, aditi hilbert wrote:
> Good point! Doing that now :)
> 
> > On Nov 17, 2015, at 9:47 AM, P. Taylor Goetz  wrote:
> > 
> > Cool!
> > 
> > For the purpose of discussion, it might be a good idea to annotate the PDF 
> > to give each one a number for easier reference. Otherwise we’ll have to 
> > reference them by description (e.g. “The green one with the ‘m’ inside the 
> > newt’s tail.” ;) )
> > 
> > -Taylor
> > 
> > 
> >> On Nov 17, 2015, at 12:32 PM, aditi hilbert  wrote:
> >> 
> >> Thanks, all. I just created MYNEWT-12 
> >> .
> >> 
> >> aditi
> >> 
> >> 
> >>> On Nov 17, 2015, at 9:21 AM, P. Taylor Goetz  wrote:
> >>> 
> >>> Confirming that no attachments were received on your reply.
> >>> 
> >>> Looks like another route is in order. JIRA attachments might be the 
> >>> easiest route.
> >>> 
> >>> -Taylor
> >>> 
>  On Nov 17, 2015, at 12:12 PM, Marvin Humphrey  
>  wrote:
>  
>  On Tue, Nov 17, 2015 at 7:32 AM, aditi hilbert  wrote:
> > I did attach the document with 4 drawings of “mynewt” to my mail. I am 
> > going
> > to try a second time with the original drawings of 6 “mynewts". Please
> > ignore the first page with drawings of newt on it. The following two 
> > pages
> > should show a total of 6 drawings. Hope you can see them this time!
>  
>  It seems that this emailing list is configured to strip attachments.  As 
>  extra
>  confirmation, I've attached the Incubator's egg logo to this email; I 
>  predict
>  you won't see it, though because of gmail's quirky deduping I probably 
>  will.
>  
>  The easiest way to share an image is probably to open a MYNEWT Jira 
>  issue and
>  attach it there, or possibly just to commit the files to the site svn 
>  repo.
>  For text there are lots of other options such as paste.apache.org, Github
>  gists or pull requests, and so on, but for images there's not a perfect 
>  answer
>  that I know of.  The downside is that we won't be able to view them in 
>  our
>  browsers without downloading first because of MIME type issues (unless 
>  you
>  commit to svn and fiddle with the svn properties), but since the target
>  audience is developers, it's not unreasonable to ask us to figure things 
>  out.
>  
>  Marvin Humphrey
> >>> 
> >> 
> > 
> 

Re: incubator-mynewt-larva git commit: add task info display. re-arrange the task structure a little, removing enum on task state and making it a uint8

[Sterling Hughes]

Hey,

Two things with this commit I wanted to see how people felt about.

1- I've done some re-arrangement of the task structure.

Specifically:

-uint8_t t_pad[2];
+uint8_t t_state;
+uint8_t t_pad;

 char *t_name;
 os_task_func_t t_func;
@@ -57,8 +58,9 @@ struct os_task {

 struct os_sanity_check t_sanity_check;

-os_task_state_t t_state;
 os_time_t t_next_wakeup;
+os_time_t t_run_time;
+uint32_t t_ctx_sw_cnt;


Prior to these changes task state (SLEEP, RDY) was an enum, taking up
4-bytes!  Bad!  (I think I may have done it this way originally :-)

If you look at this section of the structure, you now have:

  uint16_t t_stacksize;
  uint16_t t_flags;

  uint8_t t_taskid;
  uint8_t t_prio;
  uint8_t t_state;
  uint8_t t_pad;

With only 2-bits of flags being used.  I wonder if we should compress
this further, we could, for example:

- Make stack size a multiple of 8, and use 13 bits for it.  Take 2 of
those bits and make that the flags, and 1 bit to indicate Sleep or
ready.

- Eliminate state & pad.

Talking with Will, we decided this wasn't worth it.  We'd rather just
burn the 4-bytes in this structure.

2- In adding task profiling, I've made the structure a further 8-bytes
bigger, by adding context switch count and total run time.  In other
OS's, I've seen these two variable's #ifdef'd out, depending on
profiling level.

Some notes here:

- I've added them in os_arch_ctx_sw(), not within the ISR context.
This means that I could spuriously count a context switch, where the
OS wants to switch context, but before the PEND_SV (Cortex M* series)
interrupt actually swaps in the task, another, higher priority
interrupt executes and switches context to a new task.

I thought this was fine, because even if PEND_SV does switch context,
the interrupt could come before the task executed a single
instruction: there will always be some opportunity for jitter here.
I'd rather not do task accounting within an interrupt context, and
live with this.

- Total run time only increments at the granularity of the OS timer
(currently 1ms on ARM), which means that if you have smaller runtimes,
they won't be counted.  We have a more granular timer in the HAL, but
I decided it wasn't worth bringing in that dependency.  Therefore some
tasks, which execute very quickly, will show 0 runtime.

- Context switch count and total run time are very similar metrics.
However, because CSW count provides an indication of whether a task is
actually being scheduled, and runtime gives you an idea of how long
its been running, I think both are useful.

- There is one more statistic I want to add by default, which is total
stack usage.  This should be the maximum amount of stack used, as
opposed to what the current stack usage is.  I wonder if the task
should also record the bottom of the stack, so we can get the current
amount of stack usage?

Are there any other statistics we should be keeping?  Do folks think
that we should #ifdef these elements out?   I'm somewhat against it,
because I just think its useful enough to always know this, that the
slight efficiency gains you get aren't really worth it.  However, I
could be swayed.

Sterling











On Tue, Nov 17, 2015 at 10:39 AM,   wrote:
> Repository: incubator-mynewt-larva
> Updated Branches:
>   refs/heads/master 55082ce6c -> e96c2c3c6
>
>
> add task info display.  re-arrange the task structure a little, removing enum 
> on task state and making it a uint8
>
>
> Project: http://git-wip-us.apache.org/repos/asf/incubator-mynewt-larva/repo
> Commit: 
> http://git-wip-us.apache.org/repos/asf/incubator-mynewt-larva/commit/e96c2c3c
> Tree: 
> http://git-wip-us.apache.org/repos/asf/incubator-mynewt-larva/tree/e96c2c3c
> Diff: 
> http://git-wip-us.apache.org/repos/asf/incubator-mynewt-larva/diff/e96c2c3c
>
> Branch: refs/heads/master
> Commit: e96c2c3c6377452d9f7b40915f741116c8209eb3
> Parents: 55082ce
> Author: Sterling Hughes 
> Authored: Tue Nov 17 10:39:01 2015 -0800
> Committer: Sterling Hughes 
> Committed: Tue Nov 17 10:39:12 2015 -0800
>
> --
>  libs/os/egg.yml  |   5 +
>  libs/os/include/os/os.h  |   3 +
>  libs/os/include/os/os_sched.h|   4 +
>  libs/os/include/os/os_task.h |   7 +-
>  libs/os/src/arch/cortex_m4/os_arch_arm.c |   4 +
>  libs/os/src/arch/sim/os_arch_sim.c   |   4 +
>  libs/os/src/os.c |   3 +
>  libs/os/src/os_info.c| 167 ++
>  libs/os/src/os_sched.c   |  56 -
>  libs/os/src/os_task.c|   6 +
>  libs/os/src/os_time.c|   2 +-
>  libs/util/src/stats.c|   9 ++
>  12 files changed, 266 insertions(+), 4 deletions(-)
> --
>
>
> 

Re: Mynewt logo - request for input

[Marvin Humphrey]

On Tue, Nov 17, 2015 at 4:34 PM, Sterling Hughes  wrote:
> I think that is good, singular case looks better to me.  what i
> especially like about #1 is that it takes the y, and makes it look
> cool..

And you can be one of those super-annoying projects whose official
name is all lower case and induces indecision every time that the name
begins a sentence.

But if there were ever a project name where all-lower-case was justified...

Choice #1 is appealing all right.  It's a nice logo.

Marvin Humphrey

Re: Mynewt logo - request for input

[Sterling Hughes]

I think that is good, singular case looks better to me.  what i
especially like about #1 is that it takes the y, and makes it look
cool..

On Tue, Nov 17, 2015 at 4:32 PM, aditi hilbert  wrote:
> It will be hard to capitalize the M on the first one but maybe we settle for 
> “mynewt” and “mynewt OS” and change all reference to that (e.g. docs, website 
> etc.)?
>
> aditi
>
>> On Nov 17, 2015, at 3:33 PM, Sterling Hughes  wrote:
>>
>> I like the one on the upper left hand corner of the second page as well.
>>
>> On Tue, Nov 17, 2015 at 3:31 PM, will sanfilippo  wrote:
>>> I like #1 too.
>>>
>>> Will
 On Nov 17, 2015, at 10:13 AM, Christopher Collins  wrote:

 I think they are all really good, but my preference is for 1 or 2.

 I lke 1's font the best, and I like the way the newt's tail wraps around
 the y.  So actually I guess I like 1 the best :).

 Chris



 On Tue, Nov 17, 2015 at 09:49:24AM -0800, aditi hilbert wrote:
> Good point! Doing that now :)
>
>> On Nov 17, 2015, at 9:47 AM, P. Taylor Goetz  wrote:
>>
>> Cool!
>>
>> For the purpose of discussion, it might be a good idea to annotate the 
>> PDF to give each one a number for easier reference. Otherwise we’ll have 
>> to reference them by description (e.g. “The green one with the ‘m’ 
>> inside the newt’s tail.” ;) )
>>
>> -Taylor
>>
>>
>>> On Nov 17, 2015, at 12:32 PM, aditi hilbert  wrote:
>>>
>>> Thanks, all. I just created MYNEWT-12 
>>> .
>>>
>>> aditi
>>>
>>>
 On Nov 17, 2015, at 9:21 AM, P. Taylor Goetz  wrote:

 Confirming that no attachments were received on your reply.

 Looks like another route is in order. JIRA attachments might be the 
 easiest route.

 -Taylor

> On Nov 17, 2015, at 12:12 PM, Marvin Humphrey 
>  wrote:
>
> On Tue, Nov 17, 2015 at 7:32 AM, aditi hilbert  
> wrote:
>> I did attach the document with 4 drawings of “mynewt” to my mail. I 
>> am going
>> to try a second time with the original drawings of 6 “mynewts". 
>> Please
>> ignore the first page with drawings of newt on it. The following two 
>> pages
>> should show a total of 6 drawings. Hope you can see them this time!
>
> It seems that this emailing list is configured to strip attachments.  
> As extra
> confirmation, I've attached the Incubator's egg logo to this email; I 
> predict
> you won't see it, though because of gmail's quirky deduping I 
> probably will.
>
> The easiest way to share an image is probably to open a MYNEWT Jira 
> issue and
> attach it there, or possibly just to commit the files to the site svn 
> repo.
> For text there are lots of other options such as paste.apache.org, 
> Github
> gists or pull requests, and so on, but for images there's not a 
> perfect answer
> that I know of.  The downside is that we won't be able to view them 
> in our
> browsers without downloading first because of MIME type issues 
> (unless you
> commit to svn and fiddle with the svn properties), but since the 
> target
> audience is developers, it's not unreasonable to ask us to figure 
> things out.
>
> Marvin Humphrey

>>>
>>
>
>>>
>

Re: Mynewt logo - request for input

[Greg Stein]

"here are 4 options" ... forgot some attachments?

On Tue, Nov 17, 2015 at 1:28 AM, aditi hilbert  wrote:

> Hi everyone,
>
> I’d like to get everyone's input on the Mynewt logo for the Apache
> incubator website. Obviously the name is a playful play on the word
> “minute” for a small yet flexible RTOS. We are building the project
> vocabulary around the newt world as well with concepts such as egg, larva,
> clutch, nest etc. Perhaps there’ll even be an eft in the future!
>
> Here are 4 options for your review. Do you have a strong preference for
> any of them? Do you dislike all of them? If you dislike all of them can you
> please say what you’d like to see the logo capture?
>
> I also think we should capitalize the first letter (M for Mynewt) to
> reflect how we name the project now. Please weigh in on whether we should
> do that.
>
> thanks,
> aditi
>
>
>
>
>
>