KStars v3.5.0 Release Date?

Hy Murveit murveit at gmail.com
Fri Nov 13 07:12:49 GMT 2020 

Eric,

*Profile Storage*: Having those profiles in a more standard storage is fine
by me. That's a discussion with Rob--perhaps it's due to the
separation between KStars and StellarSolver?

*Re the focus profile*: sorry to hear that. Can you share an image or some
insight as to what the problem is? Ideally we could make one profile that
worked for both of us, but if we need to, I suppose we could make a few.

*Editorial comment*: Really the focus issue you bring up is representative
of the problem that we don't have a good data set for testing things like
that (I'd love 100 focus sessions from many different user setups) so
unfortunately, I design by "what seems to work for me" and then look for
feedback from the community. I'm not even sure what I'd ask for if I could
get users to send me data, but I suppose I could start with images from a
few focus sessions.

Hy


On Thu, Nov 12, 2020 at 11:01 PM Eric Dejouhanet <eric.dejouhanet at gmail.com>
wrote:

> Thanks for those fixes! On my side I adjusted the Focus UI toggling the
> capture button too soon, before detection is over. I'll publish the change
> in a few hours, and that should fix the relevant test too.
>
> Hy, if end-users now see the real value of initKeep, to me this is
> acceptable as is, even if they need to edit the profiles. However, I would
> prefer to have the default profiles read-only in a resource file instead of
> pasted in the .local folder, so that we may adjust and optimise them
> throughout 3.5.x. After all, in my own personal, own and singular (but
> probably not isolated) case, the default profile detects nothing in the
> Focus module. But I know I'm late to that party, sorry about that.
>
> eric.dejouhanet at gmail.com - https://astronomy.dejouha.net
> *De:* murveit at gmail.com
> *Envoyé:* 13 novembre 2020 07:01
> *À:* rlancaste at gmail.com
> *Répondre à:* hy at murveit.com
> *Cc:* hy at murveit.com; eric.dejouhanet at gmail.com; mutlaqja at ikarustech.com;
> sterne-jaeger at openfuture.de; kstars-devel at kde.org
> *Objet:* Re: KStars v3.5.0 Release Date?
>
> The program continues to run nicely for me in testing.
> I tested Jasem's speedup to the polar align on my RPi and worked great.
> The meridian flip checkbox issue is fixed, as far as I can tell.
> For me, the star detection, guiding and focus are all good.
> The bugs I found previously are all fixed, with a slight issue in the
> initialKeep thing.
>
> The initialKeep thing:
>
>    - Rob has submitted (thanks Rob) an MR
>    https://invent.kde.org/education/kstars/-/merge_requests/132 to fix
>    the initialKeep issue--that edited profiles lose initialKeep, and it gets
>    set to 1000000 causing possible very long computation times.
>    - I briefly tested the MR, and it (mostly--see below) looks fine to
>    me. My only comment is to add some more flavor to the mouse-over doc, and
>    ultimately to the handbook.
>    - The one issue I found, is that folks with old saved profiles, made
>    before this fix -- so all existing saved profiles as of today -- will
>    still have the problem until the user either deletes all their
>    ~/.local/share/kstars/Saved*.ini files, or explicitly edits all their saved
>    profiles and changes the initialKeep value from 1000000 to a reasonable
>    number.  Our choice is to
>    -
>       - either email out telling people to delete those files, and live
>       with complaints from a few folks who don't delete them
>       - or change the names of those files (so all people who've edited
>       profiles lose their profile work--but have a record of their lost data)
>       - I prefer the 2nd option--change the file names we look for--so
>       effectively all files are deleted.
>
> Other than the initialKeep discussion above, I'm pretty happy with the
> code.
> Hy
>
> On Thu, Nov 12, 2020 at 1:22 PM Robert Lancaster <rlancaste at gmail.com>
> wrote:
>
>> Oops. "There was also a bug where Guide always used the index from
>> focus, not it's own.”
>>
>> Thanks for fixing that!
>>
>> On Nov 12, 2020, at 3:48 PM, Hy Murveit <murveit at gmail.com> wrote:
>>
>> I found the bug related to the system always setting the users profile to
>> profile #0 whenever the profile editor is used.
>> This affected Guide, Focus & HFR, but not align (which has a different
>> implementation than the rest, it seems).
>> There was also a bug where Guide always used the index from focus, not
>> it's own.
>>
>> It was a simple change, as you can see in the MR
>> https://invent.kde.org/education/kstars/-/merge_requests/131
>> and I tagged it as 3.5.0, but of course I'll leave it to Jasem to decide
>> if it should be there.
>>
>> Hy
>>
>> On Thu, Nov 12, 2020 at 7:43 AM Robert Lancaster <rlancaste at gmail.com>
>> wrote:
>>
>>> Hi Eric,
>>>
>>> Yes it is mostly in KStars, the editor is in KStars, the default
>>> profiles are now in KStars because we wanted everyone to have more control
>>> over the defaults so they can be tweaked in nicely.  But the reason that
>>> StellarSolver affected the saving of the profile here was because the ==
>>> operator for comparing the profiles is defined in parameters.cpp and
>>> the function for conversion to and from an options map for saving or other
>>> uses is in parameters.cpp in StellarSolver.  I also have a convenience
>>> function for saving and loading profiles from a file in StellarSolver (but
>>> that was not affected by this issue).
>>>
>>> On Nov 12, 2020, at 9:10 AM, Eric Dejouhanet <eric.dejouhanet at gmail.com>
>>> wrote:
>>>
>>> Sorry if I was convinced the profile management was actually located in
>>> kstars and not in stellarsolver.
>>>
>>> eric.dejouhanet at gmail.com - https://astronomy.dejouha.net
>>> *De:* rlancaste at gmail.com
>>> *Envoyé:* 12 novembre 2020 13:18
>>> *À:* hy at murveit.com
>>> *Cc:* eric.dejouhanet at gmail.com; mutlaqja at ikarustech.com;
>>> sterne-jaeger at openfuture.de; kstars-devel at kde.org
>>> *Objet:* Re: KStars v3.5.0 Release Date?
>>>
>>> So the initial keep parameter not being saved was a bug in StellarSolver
>>> before, but I did fix it yesterday in StellarSolver before I sent those
>>> messages last night.  You might have missed it but I did mention that I
>>> fixed it with the other stuff I said.  So if you were using the older
>>> version yes it was not being saved, but now it should.  It was fixed in
>>> this commit.
>>>
>>>
>>> https://github.com/rlancaste/stellarsolver/commit/88135877f588c977e190977afca1c5132aacbbe2
>>>
>>> It was not being saved previously because we had added the initial keep
>>> parameter but then hadn’t added it to the equals method nor to the options
>>> map method used in the file saving.  Sorry I did not do this before, but I
>>> think what happened was that we were experimenting with the parameter, we
>>> were debating how it could be used, similarly to how we were in the emails
>>> yesterday, but we were still thinking about it and hadn’t yet decided
>>> whether to leave it alone or change it, but then we moved on to other
>>> things and left it unfinished.  But I did fix it yesterday after your email
>>> flagging this issue Eric and Hy’s response which reminded me we hadn’t
>>> finished it.
>>>
>>> All that remains here is to add the parameter to the options profile
>>> editor in KStars now, which I was waiting to do until we discussed whether
>>> to keep it as is or change the names.  I can add that to KStars next.
>>>
>>> Sent from my iPhone
>>>
>>> On Nov 12, 2020, at 4:05 AM, Hy Murveit <murveit at gmail.com> wrote:
>>>
>>> 
>>> Eric,
>>>
>>> Not sure which issue your were confirming.  The possible initialKeep bug
>>> would  not be minor, if it's happening it could really make processing time
>>> long for hfr and focus, like the 20s you're seeing. I'd vote to fix that if
>>> it's broken. The other one (spinbox) is an annoyance, I'd lean towards
>>> fixing it now, but if you twisted my arm I guess I could agree to wait.
>>>
>>> Hy
>>>
>>> On Thu, Nov 12, 2020, 12:56 AM Eric Dejouhanet <
>>> eric.dejouhanet at gmail.com> wrote:
>>>
>>>> I confirm, but this is probably a minor that can be addressed in 3.5.1.
>>>>
>>>> eric.dejouhanet at gmail.com - https://astronomy.dejouha.net
>>>> *De:* murveit at gmail.com
>>>> *Envoyé:* 12 novembre 2020 09:37
>>>> *À:* eric.dejouhanet at gmail.com
>>>> *Répondre à:* hy at murveit.com
>>>> *Cc:* mutlaqja at ikarustech.com; hy at murveit.com; rlancaste at gmail.com;
>>>> sterne-jaeger at openfuture.de; kstars-devel at kde.org
>>>> *Objet:* Re: KStars v3.5.0 Release Date?
>>>>
>>>> *Eric:* I see. So, can you check to see you don't have a file saved
>>>> that's messing things up (e.g. delete your ~/.local/share/kstars/
>>>> SavedFocusProfiles.ini <http://savedfocusprofiles.ini/> file if it
>>>> exists, and then make sure you're using the "1-Focus Default" profile). You
>>>> could play with the initialKeep and keepNum values in the
>>>> kstars/ekos/auxiliary/stellarsolverprofile.cpp file in the Focus
>>>> method. (Also see the note to Rob below).
>>>>
>>>> *Rob:* could there be a bug that if you have a saved .ini file, then
>>>> the initialKeep parameter no longer gets set and uses a very high default?
>>>> I wouldn't be surprised if that's what messing Eric up.
>>>>
>>>> Hy
>>>>
>>>> On Thu, Nov 12, 2020 at 12:24 AM Hy Murveit <murveit at gmail.com> wrote:
>>>>
>>>>> Also, for the record, I do agree that Rob should either expose the
>>>>> initialKeep parameter or make it a multiple of the Keep parameter.
>>>>> As it currently stands, it effectively limits the maximum value of the
>>>>> keep parameter.
>>>>> H
>>>>>
>>>>> On Thu, Nov 12, 2020 at 12:19 AM Hy Murveit <murveit at gmail.com> wrote:
>>>>>
>>>>>> Jasem: I can go either way I suppose, but in my experience, it's
>>>>>> running reasonably quickly. I won't object too strongly if you want to fix
>>>>>> it.
>>>>>>
>>>>>> Eric:  When I processed your images on my RPi4, they processed
>>>>>> reasonably quickly:
>>>>>>
>>>>>> *Looking at the 1x1 image:*
>>>>>> *Computing the HFR took 1-2s, but that's using my "quick HFR" which
>>>>>> only looks at stars in the middle 25% of the image.*
>>>>>> *Turning off the quick HFR option slowed it down to 4-5 seconds
>>>>>> (again possibly alongside guiding).*
>>>>>> *Are you using something slower than a RPi4? Could there be something
>>>>>> else slowing down the computations?*
>>>>>>
>>>>>> *Looking at the 2x2 image:*
>>>>>> *Seemed to take 1-2s without quickhfr, and 1s or less with quickhfr.*
>>>>>> *FWIW, I run my focus with 2x2 binning of a ZWO ASI 1600.*
>>>>>>
>>>>>>
>>>>>> We should figure out what the issue is.
>>>>>> Hy
>>>>>>
>>>>>>
>>>>>>
>>>>>> On Wed, Nov 11, 2020 at 11:50 PM Eric Dejouhanet <
>>>>>> eric.dejouhanet at gmail.com> wrote:
>>>>>>
>>>>>>> Which is my case right now. But in the particular subject of 3.5, I
>>>>>>> agree. I would like to know if the profiles are editable by the end-user
>>>>>>> when they are saved locally, I mean, are all parameters from the file
>>>>>>> loaded back properly?
>>>>>>>
>>>>>>> You know what I also mean here. Are there any tests validating that
>>>>>>> part?
>>>>>>>
>>>>>>> eric.dejouhanet at gmail.com - https://astronomy.dejouha.net
>>>>>>> *De:* mutlaqja at ikarustech.com
>>>>>>> *Envoyé:* 12 novembre 2020 08:46
>>>>>>> *À:* hy at murveit.com
>>>>>>> *Cc:* rlancaste at gmail.com; eric.dejouhanet at gmail.com;
>>>>>>> sterne-jaeger at openfuture.de; kstars-devel at kde.org
>>>>>>> *Objet:* Re: KStars v3.5.0 Release Date?
>>>>>>>
>>>>>>> That's fine with me. As long as it does end up generating too many
>>>>>>> stars that ends up clogging HFR calculations unnecessarily.
>>>>>>>
>>>>>>> --
>>>>>>> Best Regards,
>>>>>>> Jasem Mutlaq
>>>>>>>
>>>>>>>
>>>>>>>
>>>>>>> On Thu, Nov 12, 2020 at 10:35 AM Hy Murveit <murveit at gmail.com>
>>>>>>> wrote:
>>>>>>>
>>>>>>>> Jasem,
>>>>>>>>
>>>>>>>> This close to the release, I'm inclined to be conservative here and
>>>>>>>> let it be.
>>>>>>>> It's just something I came up with, not some reference algorithm,
>>>>>>>> and it seems to be working as it was implemented.
>>>>>>>> I suggest that we can play with this in 3.5.1 if you want, but not
>>>>>>>> mess with this now.
>>>>>>>>
>>>>>>>> Hy
>>>>>>>>
>>>>>>>>
>>>>>>>> On Wed, Nov 11, 2020 at 11:11 PM Jasem Mutlaq <
>>>>>>>> mutlaqja at ikarustech.com> wrote:
>>>>>>>>
>>>>>>>>> Hello Robert,
>>>>>>>>>
>>>>>>>>> Good catch on the partition & keep stars. I think we ought to
>>>>>>>>> resolve this not by simply dividing by the number of chunks as this might
>>>>>>>>> skew the results. In some images, some regions are more star-rich than
>>>>>>>>> others. Maybe we should do a POST star detection but PRE star filter step
>>>>>>>>> where the # of stars are then trimmed to the required initial keep?
>>>>>>>>>
>>>>>>>>> --
>>>>>>>>> Best Regards,
>>>>>>>>> Jasem Mutlaq
>>>>>>>>>
>>>>>>>>>
>>>>>>>>>
>>>>>>>>> On Thu, Nov 12, 2020 at 9:41 AM Robert Lancaster <
>>>>>>>>> rlancaste at gmail.com> wrote:
>>>>>>>>>
>>>>>>>>>> Oh the more significant question that I asked though, I don’t
>>>>>>>>>> think we addressed it yet.  Right now initial keep doesn’t work the way you
>>>>>>>>>> meant it to due to the partitions.  Does that need to be changed?
>>>>>>>>>>
>>>>>>>>>> Sent from my iPhone
>>>>>>>>>>
>>>>>>>>>> > On Nov 12, 2020, at 1:37 AM, Robert Lancaster <
>>>>>>>>>> rlancaste at gmail.com> wrote:
>>>>>>>>>> >
>>>>>>>>>> > Yep they are the same kind of thing when it comes to stars
>>>>>>>>>> certainly.  Your argument that HFR should correlate to magnitude is
>>>>>>>>>> probably very true for stars, but not for nebulae or galaxies.  They can be
>>>>>>>>>> large but dim.  It also might not be true for some stars with dust around
>>>>>>>>>> them
>>>>>>>>>> >
>>>>>>>>>> > Sent from my iPhone
>>>>>>>>>> >
>>>>>>>>>> >> On Nov 12, 2020, at 1:24 AM, Hy Murveit <murveit at gmail.com>
>>>>>>>>>> wrote:
>>>>>>>>>> >>
>>>>>>>>>>
>>>>>>>>>
>>>
>>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mail.kde.org/pipermail/kstars-devel/attachments/20201112/30b24ffd/attachment-0001.htm>

More information about the Kstars-devel mailing list