Release of v0.17

Here's the place for discussion related to coding in FreeCAD, C++ or Python. Design, interfaces and structures.
Forum rules
Be nice to others! Respect the FreeCAD code of conduct!
User avatar
NormandC
Veteran
Posts: 18587
Joined: Sat Feb 06, 2010 9:52 pm
Location: Québec, Canada

Re: Release of v0.17

Post by NormandC »

kkremitzki wrote: Sun Jan 07, 2018 6:32 pm I'm trying to make it nicer by utilizing Debian's new GitLab instance at salsa.debian.org, which means learning the git-buildpackage tool gbp and revising the package to use it (via https://salsa.debian.org/kkremitzki-guest/opencascade) but hopefully the package will be available in the next 2 weeks.
Is cloning a complete copy of the OCCT Git including history on the Debian Git really a prerequisite to building packages now? I would say that this will require a lot of space, and the Debian servers have always struggled.

The Debian packaging process has always been to create a Git repo specifically dedicated to Debian packaging, first uploading a source tarball as first commit in an "upstream" branch. The upstream history is irrelevant as far as I know. See for example the Git repo for the freecad package: https://salsa.debian.org/science-team/freecad

P.S. I didn't know Debian had switched to GitLab. I hope it will be more robust and fast that the old cgit front end (Alioth/anonscm.debian.org) which was most of the time unbearably slow.
User avatar
NormandC
Veteran
Posts: 18587
Joined: Sat Feb 06, 2010 9:52 pm
Location: Québec, Canada

Re: Release of v0.17

Post by NormandC »

Ok, I've read part of the doc for git-buildpackage and I see you are using one of gbp's supported import options. Still, I'm surprised that Debian is going along with it, at the scale of their entire repositories this represents a massive amount of storage.
jmaustpc
Veteran
Posts: 11207
Joined: Tue Jul 26, 2011 6:28 am
Location: Australia

Re: Release of v0.17

Post by jmaustpc »

This might be better discussed in another topic, however I'll post here for now......

I think we should look at what Windows versions we support with our pre-compiled downloads. If not now then definitely for the next release, i.e. 0.17.

I know we have discussed this earlier but the recent high publicity around the two new massive hardware security faults, (MeltDown and Spectre) found in mostly Intel 64bit CPUs (apparently nearly all of them since 1995!) as well as some ARM CPUs, made me wonder if we should reassess if we are being irresponsible supporting Windows versions that have had all MS support dropped. Mostly due to the PPA system, we have more or less automatically dropped support for old unsupported Ubuntu variants.

This is a link to one random article about the above "Meltdown and Spectre".
https://www.theverge.com/2018/1/3/16844 ... dows-linux

XP was used on some CNC machines. I saw a CNC machine running Windows many years ago, I am sure it was before XP was ever released so I think it had Win98 or even 95. To the best of my knowledge, no one was stupid enough to offer CNC machines running Vista, but I don't claim to be an expert on any of this.

In reading about Spectre and Meltdown, I noticed that MS will not be providing any patches to protect Vista users. I was wonder why as I though MS was still supporting Vista but apparently they dropped all support in April 2017! Here is a random article about the dropping of Vista support....(by the way MS's solution is to buy a new Windows and they suggest win10 which I thought had famously dropped support for many not so old CPUs etc.)..
https://www.cnbc.com/2017/04/11/windows ... osoft.html
Obviously a better solution is to install Linux, probably Ubuntu or a Ubuntu variant.

This is an article from 2009 about the then upcoming Windows 7 regarding 32bit and 64bit versions.
https://www.techspot.com/guides/177-win ... bit-64bit/

I would have like to say also lets drop Win32 but since Win7-32bit is still supported and there may well be a lot of 32bit users out there, we shouldn't drop 32bit yet. I guess and could be wrong, that many Win32bit downloads are mistakes and are being run on Win64.

Anyway I suggest that
1) (this is less important) we deprecate 32 bit, still make it available but make it a little more difficult to find and add a warning of some sort.
2) we do not offer any pre-compiled binaries for download which support Windows earlier than Win7

If you guys want to discuss this in any detail then I will happily spilt it off into another topic.

Jim

Edit: - To give you some idea of how low the Linux Kernel developers opinion of Intel letting this bug exist is, read this article at "The Register" linked below and note about halfway down under the heading "impact" that the Linux guys were going to produce a patch named "Forcefully Unmap Complete Kernel With Interrupt Trampolines" so that its acronym could be "word censored".
https://www.theregister.co.uk/2018/01/0 ... sign_flaw/
chrisb
Veteran
Posts: 53930
Joined: Tue Mar 17, 2015 9:14 am

Re: Release of v0.17

Post by chrisb »

I think this could well be split off, because it seems to be independent of the next release, although that would probably be a welcome occasion to change other things too.

I appreciate to have the standard downloads on some main page and to have further downloads on a secondary page. There sure will be some requests for 32Bits or even for a then good old 0.16.

But I don't think it is our business to not support insecure operating systems. It is an opportunity to drop support if it is too much additional effort and of course FreeCAD must not use any security holes in order to spy personal data, but it is solely a user's business where and in which environment he uses FreeCAD.
A Sketcher Lecture with in-depth information is available in English, auf Deutsch, en français, en español.
User avatar
saso
Veteran
Posts: 1920
Joined: Fri May 16, 2014 1:14 pm
Contact:

Re: Release of v0.17

Post by saso »

chrisb wrote: Thu Jan 11, 2018 7:45 am But I don't think it is our business to not support insecure operating systems. It is an opportunity to drop support if it is too much additional effort and of course FreeCAD must not use any security holes in order to spy personal data, but it is solely a user's business where and in which environment he uses FreeCAD.
+1
abdullah
Veteran
Posts: 4935
Joined: Sun May 04, 2014 3:16 pm
Contact:

Re: Release of v0.17

Post by abdullah »

wmayer wrote: Fri Jan 05, 2018 2:56 pm
Must be this one issue #0003084
It is on my name, I will try to fix it. It should not be too difficult.

Do you have any other bug you would like me to take care of?
wmayer wrote: Fri Jan 05, 2018 2:56 pm
abdullah wrote:...
OK, I will look at these PRs first.
I am back to bug fixing. It would be great if this is not delayed further, so that there is time to fix potential bugs.
abdullah
Veteran
Posts: 4935
Joined: Sun May 04, 2014 3:16 pm
Contact:

Re: Release of v0.17

Post by abdullah »

chrisb wrote: Thu Jan 11, 2018 7:45 am But I don't think it is our business to not support insecure operating systems. It is an opportunity to drop support if it is too much additional effort and of course FreeCAD must not use any security holes in order to spy personal data, but it is solely a user's business where and in which environment he uses FreeCAD.
+1

Two reasons:
1) I can indeed imagine people on old computers not connected to the Internet and with very limited contact with infectious software. The world is very big.
2) I do not use windows and I am not a windows fan, but if something I remember from Windows made me extremely angry is that it is not only that they drop support (it is a company and they have to make money), but they drag everybody else to drop support..., so if not too much additional effort, I would be delighted to see FreeCAD supporting them :twisted: :twisted: :twisted:

On the other side, I encourage everybody still living in the dark side to embrace the opportunity to run a real OS. Mint, Ubuntu, Fedora... they are nice and user friendly flavors of linux. Your computer will run faster (unless obsolescence programmed already in the firmware) ;)
User avatar
DeepSOIC
Veteran
Posts: 7896
Joined: Fri Aug 29, 2014 12:45 am
Location: used to be Saint-Petersburg, Russia

Re: Release of v0.17

Post by DeepSOIC »

abdullah wrote: Thu Jan 11, 2018 10:22 am Do you have any other bug you would like me to take care of?
There is a bug with endpoint tangent constraint on bsplines. It sometimes wants to switch direction when poles are dragged around. Although I would consider it a high impact bug, it is not that important, since creating such a constraint is not straightforward (requires deleting weight circles).

I'll try to make a bug report today. I have experienced it in a rather convoluted project, and I was too busy with getting it done back then, I couldn't afford spending time on nailing the bug conditions.
chrisb
Veteran
Posts: 53930
Joined: Tue Mar 17, 2015 9:14 am

Re: Release of v0.17

Post by chrisb »

DeepSOIC wrote: Thu Jan 11, 2018 12:58 pm There is a bug with endpoint tangent constraint on bsplines. It sometimes wants to switch direction when poles are dragged around.
Is that different from the situation with circles and tangents? It seems to me to be the uncomfortable but well known difference between a sketch having 0 degrees of freedom and being unique.
A Sketcher Lecture with in-depth information is available in English, auf Deutsch, en français, en español.
abdullah
Veteran
Posts: 4935
Joined: Sun May 04, 2014 3:16 pm
Contact:

Re: Release of v0.17

Post by abdullah »

DeepSOIC wrote: Thu Jan 11, 2018 12:58 pm There is a bug with endpoint tangent constraint on bsplines. It sometimes wants to switch direction when poles are dragged around. Although I would consider it a high impact bug, it is not that important, since creating such a constraint is not straightforward (requires deleting weight circles).
No, it does not require to delete the pole circus circle ( polar circus is also funny :lol: ). It can be done from the element widgets. Just go to your favourite bpsline in the element's widget, press "z" to iterate between edge, starting point, endpoint. Once the point you want is highlighted, click to select that point. It is by far not the most ergonomic way (it is probably the less ergonomic way actually ;)), but it can be done. Of course it can also be done from python directly :P

About the bug, once you can reproduce, file a bug and I will take a look.
Post Reply