PixInsight 1.7 discontinued

[Juan Conejero]

Hi all,

Today we have removed all PixInsight 1.7.x installation files from our file servers. We have also disabled all updates for the 1.7 series.

Since the current version 1.8.0 RC7 of PixInsight has proven more stable, faster and more efficient than previous versions, let's move on.

[pfile]

i hope you are right, i still get the occasional "hard crash" out of 1.8RC7 with no backtrace or anything (osx 10.8.4)  hard to debug that.

[bitli]

It is nice to move on, and I agree than 1.8RC7 is very stable.

However from a customer point of view, I would prefer to see an announce like: 'as 1.8RC7 is very stable, we will make it our main release and retire the version 1.7 in 2 weeks'.  Or maybe even better: 1.8RC7 is now available as 1.8.0, but is identical as 1.8RC7 (so that new people do not have to wonder why there are only RC of PI - after all 'RC' is suppoed to have a meaning).

It would not change anything materially, I would just feel more comfortable.

This being said, I recognize that there are other reasons why you may have to act so quickly, like license broken or anything you may not want to discuss publicly.
--bitli

[darkownt]

I agree that it is odd to only have a "candidate" offering or equivalently to discontinue supporting the only actual "release" available.

In some sense it is more honest than what other software companies have been known to do: providing no support for existing software products or releasing as products versions that aren't really ready (very unstable).

If things ARE in limbo... I guess it's better to know than to be mislead.  Still seems odd.

[Carlos Milovic]

There are going to be a few modifications to the current release, before it is launched as the official (final) 1.8 version. Most of them are going to be behind curtains, and some bugfixes, but we are also integrating new processing tools (modules). So, we ask for a little patience. Our development right now if built over 1.8RC7, and the 1.8.x cicle has officially begun with Juan's announcement.

[mschuster]

IMO sounds like what you will release next will effectively be 1.8RC8.

Mike

[Juan Conejero]

Since I am being questioned... I have discontinued version 1.7 just because I have very good reasons to do so. I prefer to not comment more about it; surely everybody can figure out.

However from a customer point of view, I would prefer to see an announce like: 'as 1.8RC7 is very stable, we will make it our main release and retire the version 1.7 in 2 weeks'.

I prefer to tell the truth. The truth is that we have a production-stable release candidate version. It is not the final 1.8.0 version mainly because it still lacks some new and improved tools, some GUI improvements and a lot of PJSR functionality that I want to have included in 1.8.0. Hopefully it will be ready in September. If this is causing a climate of distrust then I am sorry, but that's all I can offer for now.

[Peter]

Juan there is absolutely no reason to apologise, it is obvious that you wish to take all PixInsight users to the highest level and I am in total awe as to what you and the other guys give us. I am sure I am speaking for most if not all users to say a BIG thank you for this superb program. Let it continue forever

[Philippe B.]

A agree with you Juan. 100%
As 1.8 is now VERY stable (and it is really !!!) with new speed improvement, it is time to put 1.7 in the past.
When you changed from 1.6 to 1.7, there was no discussion. 1.7 has replaced the 1.6, that's all.

I don't understand why some users don't want to switch to 1.8 when you see ALL the newest features it has.

On the next PI training we will do at AIP in 2014, we will not support 1.7 nor 32 bits versions.
All trainers should have 1.8 64bits (Win, OSX, Linux) with latest release at the date of the training course.
Last year, we spent many time with guys who had 32bits version or 1.7 where they cannot access to the newest scripts ot functions or they got memory full when they integrate 10images because their computer is corrupted.... blablabla.


Next discussion will be to discontinue 32bits 



Juan, many thanks for the newest functions of this week (intergration, starAlignment...) very nice improvement ! Speed, accuracy.
We have to talk about TGVdenoise. I'm not 100% happy (only 99%  ). There should be some improvement to do there, I think.

Philippe

[Carlos Milovic]

Philippe, I would appreciate if you send those commenta to me. I'm writing the next update of tgvdenoise, so I could use your inputs.

[raf1]

Juan there is absolutely no reason to apologise, it is obvious that you wish to take all PixInsight users to the highest level and I am in total awe as to what you and the other guys give us. I am sure I am speaking for most if not all users to say a BIG thank you for this superb program. Let it continue forever

Amen

[pfile]

I don't understand why some users don't want to switch to 1.8 when you see ALL the newest features it has.

Philippe

because it still crashes for me, hard, with no backtrace, that's why.

my machine is extremely stable and will run memtest for 48+ hours with no errors. also 1.7 does not suffer this behavior.

when i use 1.8RC7 i am compulsively saving the project, like every 10 minutes. twice i have lost everything i was working on. i can't let that happen to me again.

maybe it is an OSX thing and it's my fault for using OSX. i don't know.

sorry to spoil the lovefest but i also prefer the truth.

[Sean]

Pfile,

I'm running Ripley RC7 on OSX 10.8.4 and have seen no problems at all - it's very stable. I have a number of large PI projects with lots of open images and multiple processes, and my system has with lots of other programs running at the same time.

I'm using a mid-2011 iMac with 12GB RAM, Intel Core i7, Radeon Graphics, dual monitors.

If I can help you to determine the source of your problems, please let me know. I'd be glad to help.

Sean

[pfile]

sean, just wait, it will hit you.

in my latest project PI did not want to save the project when i was done, since one or more views was locked. as an aside, when this happens, the dialog is singularly unhelpful. so, it's dangerous to save the project when a view is locked? okay, what am i supposed to do? undoubtedly a view was locked because some process left it locked, not due to anything i had done. maybe the dialog could describe some steps to determine which view is locked and how to possibly unlock it.

while working on that same project, i got a segmentation fault on some process. i was lucky to be able to save the project and quit PI in that state. while it was in that state, something was very wrong with the real-time preview - it was taking a minute or two to display anything.

as i said, the machine is totally stable. no other programs die on me. my machine uptime is regularly measured in months. as i said the machine passed burn-in running memtest86 on all 32GB of ram. while i would agree that it's insane to run that much ram without ECC protection, i can't really do anything about intel's choices. all i know is that if i had problems with bitflips, other programs would crash on me or the kernel itself would eventually crash, but this never happens.

remember when i reported to juan that OSX was cleaning up in /tmp after the swap files' atimes exceeded some scripted limit in the OS? that limit is like 3 days by default. i'd regularly have 1.7 running far, far past that horizon. with 1.8RC7, i've never had it run that long. same hardware. same OS. different version of PI.

so i'm not sure what you can do to help me debug it. i never run PI 1.8RC7 unless it's attached to a debugger, and that has saved my bacon several times. sometimes i can pop the call stack and resume execution and then immediately save the project.

let me also just point out that what we are on is called RC7 and there were some very serious, deep, and difficult-to-solve bugs in all prior RCs, especially RC6. i have no doubt that there are still some bad bugs lurking around. in all fairness some of them may have come from 1.7 and are still around, but overall 1.7 was much more stable than 1.8RC7 for me.

rob

[Grey]

A colleague from our observatory is currently testing the trial. He still hast a 32 bit OS and is not so familiar with operating systems and so on. He tries to run the stacking script, but it fails saying

"At address 5A17ED1B with exception code C0000005 :

Access violation: invalid memory read operation at address 00000000"

So my idea was switching back to 1.7 for the testing, but this is not possible any more (I used it in the past and it was fine working for me). Support for the 32 bit release is also not available. From the customer point of view, the best solution would be to keep the old stable 1.7 32 bit release available, so everybody can use PI in there current environment. Of course it makes totally sense to use 64 bit, since we can use more RAM and so on.

I am currently not sure, if it is the best way to quit 1.7 when the development of 1.8 is still running. In the end 1.8 is still a RC and it should be the decision of the user to use a RC or not (I also work in the software development business, it is not so easy to find the best solution for everybody).

Thanks,
Tobias