StarTools

First of all, let me to congratulate you for this excellent piece of software. It takes some time to get used to it, but once you get comfortable wtih, it becomes extremely powerful.

I come with something that I believe you programmers love to fix, a repeteable bug. I'm using Startools 1.3.5.283 RC 64bits Demo version but these bugs are not new to this version, simply I've discovered how to reproduce them.


Bug #1: Corrupted image. I've tested it on a pair of images and the bug is repeteable. In case you need it, here is a link to an image with wich I can reproduce the bug

https://www.dropbox.com/s/kitsu6f2of5ug ... %20Leo.fit

- Open as "Linear, was Bayered, is not whitebalanced"
- Autodev with default values
- Wipe with Temporary Autodev On and agresiveness set to 90 and darkpixels to 5 (may be changing values is enough)
- Development to 95,62% (or something useful)
- Deconvolution using the automated mask. At this point you get something like this. - Manually develop it again and the image becomes corrupted like this. Of course this is not a very logical processing sequence, but it is the way to reproduce an error wich happens frecuently if you like to refine development while processing

I've found that, after being stuck in the middle of a processing with this error, if I Restore to "Globally stretches, Wiped, Deconvolved" it restores the image even if I lose some of the work

Bug #2: Startools Crash

This is a more elusive bug, it doesn´t happen always and I think it has more to do with Startools background processing.
As before, here is an image(very poor of course) that makes Startools to crash
https://www.dropbox.com/s/84k0usd9u96t9bv/ngc1333.zip

- Open a fairly big image like 3000x2000 as "Linear, was Bayered, is not whitebalanced"
- Autodev with default values
- Enter Color module
- Mask->Auto->Stars (very noisy selection in my case) and Keep
- Press Sample
- Mask->Clear->Invert->Keep
- Inmediately press Sample and Startools crashes

The procedure is really idiotic but it is the way I've got to reproduce the crash. I hope this bug is the same that the one that really annoys me and that I've not been able to reproduce. It happens when a common procedure to calibrate color, that is, Mask->Auto->Stars->Do->Keep | Sample | Mask->Clear->Invert->Keep , when executed with no delays, crashes Startools


If you fail to reproduce these failures I'll be happy to send you more information. My system is an Intel i7 with 16Gb RAM on Win7 64bits

P.D. Thanks again for this wonderful software

Many, many thanks for the comprehensive bug report and associated data!
This is absolutely invaluable.

I'll get back to you once I have had the chance to investigate.

Thanks again!

Ok, a new version is up that yields better results when doing a restretch using AutoDev (but, alas, not yet for a manual Develop).
The reasons for this (as you may have already suspected) have to do with the Tracking information and how much of it is stored to compensate for non-commutative operations. There is extra work that I need to do in this area. While the Tracking engine has exceeded my expectations throughout the 1.3 life cycle, some redesign is in order to improve behaviour in fringe cases such as these.

I rarely use a manual Develop though, as AutoDev with an ROI almost always yields much superior results (so too with the data that you so graciously shared with me).

I am yet to replicate the crash but I will keep trying.

Again, thanks so much for this comprehensive bug report!

Thank for the attention, Ivo. I've downloaded the new 1.3.5.284 RC version and i'm not able to reproduce bug#1 whith those steps, I had to play a little bit more wih the image to get it corrupted again so, it is a somewhat good revision. I've also tried to play with the image using only AutoDevs and I got it corrupted again, but this happened doing more than 4 Autodevs between steps, so it's not a concern.

Obviously bug#2 persists. Tell me if i can help you to track down the problem. It also would be useful if others users try to replicate it and post it here.

This is the debug error in case it helps you It crashes in the same way with 32 bits version

Regards

Ok, I *finally* managed to reproduce this bug, across all versions too. It was a nasty one... (it required some specific timing)
It should (hopefully) be fixed in 1.3.285RC, up for download now.

Thanks again for reporting this as well as your patience!

I'm glad you finally nailed the bug! You are going to hate me, but you forgot to update the text in the web download link line

Edit: With this new version it seems to works fine, I've not managed to crash it

Regards

Alvinillo wrote:I'm glad you finally nailed the bug! You are going to hate me, but you forgot to update the text in the web download link line

Ugh... Brain not working today...

Edit: With this new version it seems to works fine, I've not managed to crash it

Awesome!