AbeAnd Logs March

Fri Mar 31, 2017

I was putting the nut on the bolt in my cube and adjusting sizing to what I thought it should be when freecad disapeared. It did successfully recover the freecad file at a recent point, but apparently when it crashed vokoscreen lost the video file. Probably only because I thought it more efficient to record the freecad window instead of the whole screen, now I know better. It took me ~20 mins to get to the bolt so I'll just start over and record again at least I should get faster. I think the harder part to me is still the sketcher assuming the need to create effecient and complete constraints, although unnecessary in this case.

I have got most functions I am used to using working in Linux so far, but not without significant effort and I have noticed more than a few simple bugs in the OS GUI. It has also occasionally thrown up system crash error reports for no apparent reason. So while Linux or Ubuntu has improved it still has a lot of room for improvement.

Wed Mar 29, 2017

Working with freecad assembly2 bench trying to understand optimal methods of assembly and consequences of various options. I checked out the animate function intended for looking at constraints as well. I can see how it is useful for more than that as well, but a proper animation bench is probably necessary for complex machine animation. I also started testing RayTracing with LuxRender and then POV-Ray. Both seem to have issues I am uncertain how to adjust. Luxrender seems slow and/or not lit properly. POV-Ray seems to quickly, but incorrectly render the scene geometry. I am also unsure if any OpenCL GPU acceleration is working. I just don't know how to check on Linux and the documentation for the renderers and freecad is sparse at best. Regardless, I have a good idea of how to assemble parts so I am ready to do the test.

Tues Mar 28, 2017

Watched team meeting from yesterday on YT. Interesting stuff, but I am definitely not up to date on the current state of 3d printing or the big picture of where the team is at in the build process exactly.

Signed up for Minds then realized the OSE network is a separate implementation, which I think I read before, but I am not familiar enough with the interface yet.

More FreeCAD; read related wiki pages, docs, YT tutorials etc.

Also looked at OpenSCAD function and language. Very interesting, but I can see a definite learning curve for more complex tasks. The ability to use OpenSCAD within FreeCAD could have great potential, though. FreeCAD may need more development and debugging, but from all the tutorials out there on YT there is plenty of functionality there already including animation methods and more work benches to install.

Working with assembly2 bench and preparing to build a cube.

Mon Mar 27, 2017

Reviewed FreeCAD best practices. Looking at assembly2 workbench and considering best practices for creation and drawing versus editing existing files, which will be more common in the future as the parts and machine library continues to grow.

Thurs Mar 23, 2017

More freecad tutorials. Looking for a thorough one on assembly. I might need to do more research and make one. Finding the optimal order of operations for working and computation efficiency seems important with freecad. I'm sure many improvements in code are coming, but currently, the method of breaking down components into smaller parts and grouping later helps both speed of production through distribution and saves processing time for users especially with slower machines. With the complexity of assembly being handled by more experienced users with likely better hardware program efficiency is less of an issue. Training users to use the software efficiently will help with productivity though so understanding and explaining freecad's underlying functions can help speed design.

Wed Mar 22, 2017

Looked over freecad scripts page, docs, & worked through tutorials. Earlier, I was thinking it would be nice to make outlines for creating tutorial videos so this is great. It seems better to script important tasks for learning than just record work to cut together later. Certain functions seem best learned through practice. Seeing the process & explanation of specific demos of how functions like the solver & constraints work is helpful for learning to optimize work.

Sat Mar 18, 2017

Used a firefox to help download some media files to build a library of recent CAD and graphics documents. Still thinking about how to organize files, but I put a copy in my syncthing folder for access and testing. I've been adding a searching for extra apps and tweaks for my Ubuntu install to get it more useful and learning more command line tools as I go. Showed the 2011 TED talk to a couple WWOOFers.

Fri Mar 17, 2017

Finally got new ISO (3/14) to boot from USB first using Rufus back in windows and then Unetbootin in previous linux version. Methods using mkusb or various dd commands have not booted. Testing the persistence mode with Unetbootin has not worked. Text files made on the desktop and Documents folder are not saved despite the option being set and choosing it from the boot menu.

I tested the various MBR and UEFI methods. I'd guess the MBR both BIOS and UEFI method would be most compatible for anyone wanting to use a stick in different machines. The new Ubuntu live stick I made with Unetbootin booted farily quick the first time from the grub menu. I checked for various apps and the kdenlive icon looks different so I ran that it gave a warning about vlc or xine not being installed. Not much other apps like gimp or audacity in this version. This version is obviously smaller. Unfortunately, Rufus does not support persistence. Linux Live USB Creator is supposed to and that could be tested again as well for windows users in the future.

mkusb gives me the following warning when running from my Ubuntu installtion:

dus-persistent 12.0.3 cannot install 'grub-pc' in installed systems in UEFI mode. 'grub-pc' is is used to make persistent live drives bootable in BIOS mode.

You *can* install 'grub-pc' in all live systems, so if you want a persistent live system, that can boot in BIOS mode, you should run dus-persistent in BIOS mode or in a live or persistent live session in UEFI mode.

Wasn't sure I understood this until I read this:

https://help.ubuntu.com/community/mkusb/persistent#Using_data_from_the_source_iso_file

So I also tested making another USB from within the live USB like mkusb recommends for 'grub-pc' option, but the stick did not boot. No alloc magic error like before just a blank screen. mkusb's partitioning and boot setup seems overly complex. I am surprised dd does not work though. Still trying to understand what incompatibility dd has with new ISO's. When trying to boot sticks made with dd I generally got no OS found errors. My next guess is to mount the ISO use dd to copy and make bootable. The Rufus install log did detail information about the ISO boot methods including syslinux. However, the final log I saved looks different so I may test this again.

Rufus version: 2.12.1054 Windows version: Windows 10 64-bit (Build 14393) Syslinux versions: 4.07/2013-07-25, 6.03/2014-10-06 Grub versions: 0.4.6a, 2.02~beta3

Also testing some download tools to automate downloading files.

Wed Mar 15, 2017

Downloaded new ISO Tested using to use mkusb. mkusb ran ok, but would not boot. I thought it mucked up stick partitions from previous ext4 part & didn't set bootflag so I wiped 1st 1MB tried again it was setting boot flags. It makes multiple Partitions for boot, UEFI, NTFS for user persistence data.

Tried booting again no luck & soon system UEFI stopped booting devices. Powered off. Reconfiguring in UEFI. UEFI no longer detects any USB or SATA devices connected to system !!?? Posting from Android. I have to go move some pigs deal with this !?$#& later...

Turned off PSU earlier figured it would reset the UEFI weirdness and it did. mkusb has an interesting interface and options, but I liked the plain ext4 partition with boot flag best. I think that is most compatible. I can see the partitions for BIOS mode, UEFI, ext? for casper-rw, and NTFS User data being useful or necessary in some situations, but complexity is good to avoid. Once the UEFI was working again I tried booting the stick again in different modes UEFI gave an error "error: disk`hd0,4 not found. alloc magic is broken at 0xad1883e0: ad00de60 Aborted. Press any key to exit." Legacy USB mode couldn't detect an OS probably because the boot flag needs to be set manually on the FAT partition for that.

mkusb uses dd, but obviously, adds some unnecessary complexity. So simple dd command line may be the most consistent option.

Also testing out Syncthing and Added ID at top of log. Just reading over the Syncthing documentation it looks promising as an open source cloud sharing platform and I don't see any info about limitations on the max number of clients. https://syncthing.net/

Tues Mar 14, 2017

Creation of ISO with apps from spreadsheet list and more using cubic and shell scripts appears successful, but needs testing.

Shared scripts in gdrive, but added a WARNING about the dangers of executing publicly editable shell scripts.

Created and embeded some collaborative gdocs to help speed sharing of information beyond the wiki.

Mon Mar 13, 2017

Trials and errors giving way to progress. I quit using the VM installation and I am now running my earlier installation of OSE Linux on disk.

I figured out cubic doesn't want to copy in files with spaces in the name. After that it was easy to modify and copy in some shell scripts for install. I also see where it successfully made an ose folder in the squashfs home folder. I also looked for repos & ppa's for more software on the list. Some apps like qucs are coming up as possibly incompatible or not officially verified for 16.04, which may be another reason to go back to 14.04. There may be newer apps or competing alternatives to some of the apps listed as well like ButtleOFX versus Natron. many of the apps are fairly advanced and niche, so dev team's using them can debate and advise what should or should not be included per the quarterly release cycle.

Cubic is in the middle of apt get script downloads so I am looking at ways to download/data mine the wiki for the recent file uploads that should be included. Since the wiki is the current central repository data mining it with a web spider app like http://nutch.apache.org/ or https://scrapy.org/ might be the quickest way to sync recent files into a folder in the ISO. The web cast might be to wide and require some detailed scripting though. Once the desired older files are included mining the upload logs periodically might make the process easier.

Sun Mar 12, 2017

Testing and adding more packages via package shell scripts for working with cubic/command line ISO editing. Problems with VM install possibly caused by wireless internet and weather hosed my package system to the point I could get any package managers to resolve forced me to rebuilt the VM. Now Cubic is not loading ISO's. Other dependency issues? Nope, copies of ISO's failing MD5SUM check... re-download... back to freecad. Maybe Bit Torrent can correct the 16.04.2 ISO file quickly?

Cubic is only useful in that it lets you move back and forth through the editing process with a GUI and reduces some of the command line work with shell scripts.

Added a few ideas on IT support. Fake it till you make it approach as always. hm that could be a tagline? hack it till you make it?

Sat Mar 11, 2017

Testing Cubic did not have desired results. I thought it copied files to a folder in root, but they are nowhere to be found in the ISO, squashfs, or upon trial booting. I will try again and I think a combination of the command line methods with cubic might be ideal. The location and user accessability of the files is also important. According to one guide the trial boot user interface is generated at boot time and not editable, but it must be generated from some layout instructions so it is a question of how difficult it is to modify and maintain. All this leads me to reconsidering the method of transfering a installation image to USB without any proprietary drivers would be better for user performance, implementation, and maintenance.

Google has apparently released a new version of hangouts through G Suite enterprise, which costs money.

https://gsuite.google.com/learning-center/products/hangouts/get-started-meet/

Open Source solutions can't be to far behind.

Fri mar 10, 2017

I was about to dive into a long manual command line ISO customization attempt again when I found cubic. I spent to much time trying to organize possible concepts on Talk:OSE_Linux. I need to edit and submit faster and organize early. I am uncertain about creation and implementation options within an ISO/image affecting long-term support capability. There may be difficulty initially aggregating files from the wiki to include in an ISO. The wiki is great with versioning etc., but it would be nice to have a universal synced library folder structure with the latest files as they're edited by dev teams and to have complete archive packages for each version of a machine. Google drive only partly achieves sync with data being spread across users accounts relying on proper public sharing to sync and it isn't technically open source. It could be great if an alternative can be found for the OSE Linux platform, but I don't see the needed features in the common P2P sync apps yet.

Thurs Mar 9, 2017

I've been reading some general info on the functionality of Live boot systems and persistence with casper-rw. From what I can grok the existing ubuntu Live install media and tools like Unetbootin and Rufus are the simplest way for most users to make USB Live boot devices with persistence. While most of the file system is read only some extra space (up to 4GB for FAT32 file size limit more for ext4) can be setup to hold user files and save system state info upon shut down. There are different options for the filesystems like UnionFS, OverlayFS, and aufs (used by Knoppix for greater stability). Another option is distributing a raw image taken from a USB device pre-made using such tools and then having users use other tools to clone the image to a USB device. It would technically eliminate steps, but the tools to copy the image would need to be user friendly and functional on all platforms.

Tues Mar 7, 2017

Continued trying command-line method to edit Linux ISO by making a shell script. I keep making mistakes trying to automate the process, but I am assuming the whole process works. I'd like to find a more automated method to edit, but currently it looks like most apps for this are outdated. I am also curious about different ways of making persistence work for the Live USB mode. While it is possible to do different types of installs even from one USB to another USB that might be confusing to many users. Unetbootin describes different installers/methods of booting. I think the ext4 formatted USB device I'm seems to run better in general and had more options from the bootloader menu, but it could the hardware is better as well. Also, the superiority of ext4 verse FAT32 probably has multiple benefits if users intend to run Live USB only. Again, however, that requires an existing Linux installation or a separate GParted Live USB just to format another Linux Live USB device.

During the ISO edit process, I also realized my concerns about too many packages and size may not be a problem yet because the file system is quite compressed in the ISO. I haven't successfully recompressed a new version with more packages yet so I can't guess at the new size yet. This is probably also a hindrance on performance, except the whole system goes into RAM. Which, in turn could be a performance issue. An installer closer to a USB to USB install would likely have much better performance. I think there are different methods for executing Live USB systems and it will be worth looking at alternatives. There are some specialty Live USB Linux versions like Puppy and mint, but they are mostly stripped down light weights.

Still uncertain where and how extra files CAD library etc. should be placed in the ISO structure other than the maybe the user Documents folder. Maybe that would copy over or maybe it would require some extra modification of the installer. Since I am not a very experienced Linux user more research is necessary, but with Linux that is almost always the case.

Sun Mar 5, 2017

Successful installation of OSE Linux from bootable USB devices made using Unetbootin in VB VM running OSE Linux. Some glitches and oddities, but it worked out.

I did some different tests just to verify likely problem spots for users and get a general idea of how well different devices work. I'm still doing more tests and I'd like to get the ISO and USB process a little more polished.

I read more Ubuntu USB stick install recommendations and there is better documentation for 16.04 than I saw previously. Most of the bugs occur with differing software versions, older boot device creator software and the system BIOS/EFI/UEFI. There are also occasional USB device incompatibilities. Just getting a UEFI system to boot from USB device is not always perfect either. Pressing and tapping frantically at the boot menu key doesn't work and a restart and retry is required. This can be more of an issue on faster modern systems with fast or ultra fast boot settings. Many users may need to adjust BIOS/UEFI settings just to be able to boot an alternative device.

I tested a 16GB USB 2.0 Patriot XT device by using gparted to format ext4 and used Unetbootin to transfer the OSE Linux ISO to it. First, I left the boot flag unset I wanted to verify the EUFI would give a no boot system found error. It did, so I then set the boot flag on in Gparted. It booted fine I ran the try ubuntu option and then the ubuntu installer from there. It went smooth and reasonably quick. I also formatted an 8GB USB 3.0 (unknown brand) to fat32 and set the boot flag in gparted then again used Unetbootin. It was slower and had a few glitches after booting into try ubuntu and during the installation setup it hungup a couple times. So I rebooted it and ran the ubuntu boot menu check disc option to verify the USB installer. It passed verification and the next boot I used the Install option from the boot menu and I was able to partition and complete the install. One oddity that might be due to Unetbootin (frugal installer?) is that after booting into the full installation it still had the installer option on the launcher. So I think it did not do a proper install as much as it copied the executing system to the new disk.

For 16.04 Ubuntu recommended tools are for windows users to use Rufus and for Linux users the latest version of usb-creator-gtk. I am curious if all of these boot device creators set the boot flag or install an MBR boot sector as I recall that being an issue for me in the past, but I've learned it is easiest to use gparted to double check. I plan to test a few more boot software options and if needed edit a new ISO using command-line instructions so I can test and complete a full install with all the up to date software packages and OSE libraries so that it is ready for me to use post install (or off the stick).

Also when I booted into the ubuntu installation on my SSD the first time I saw the installer option still available I assumed I had set the UEFI boot options wrong and it booted off one of my sticks again. I rebooted checked the UEFI tried again got a GRUB minimal BASH twice restarting and checking UEFI. I turned off some boot options, but the GRUB 2 loader is what I installed to the SSD and that is what was glitching. I recall having boot loader issues like that with previous Linux installs. Not sure if the EUFI read something wrong the first time or GRUB 2 adjusted itself somehow, but patience and reboots are often required. One expects the software/hardware to execute correctly every time, but some things about BIOS/UEFI and boot loaders are clearly imperfect. The way the EUFI and GRUB handled the install overall is good, though. The EUFI automatically detected and defaulted to boot the ubuntu SSD first and the GRUB menu detects and allows selection of the other windows boot hard drive. No need to modify the original system disk and I can move the internal SSD when feasible with no software changes too the original system.

Sat Mar 4, 2017

Working on getting OSE Linux installed on new SSD I added yesterday, but in a roundabout way. I'm running the current OSE Linux ISO in a VM, testing putting it on USB sticks to test in other VM's and system UEFI boot testing, while testing Unetbootin and working through the ISO editing terminal instructions I found for 16.04.

Thurs Mar 2, 2017

Dealing with a few annoying local technical difficulties the last few days. Mainly trying to download OSE Linux ISO. Started a direct download from google drive then it failed to resume after a low bandwidth interruption. Then I decided to use the drive sync. Went back to work in my existing Virtual Box VM and it was inaccessible/corrupt for no apparent reason. My past experience with VirtualBox and Linux has been much better I am surprised how beta this supposedly non-beta software is behaving. Then Power outage. Everything was back on the next day then grid repairs sent surge spikes and brownouts and WISP was down most of the day. It will great to get off grid some day, but in this case, it wouldn't have helped much. Finally, realized I needed to uncap google drive sync a bit. Now it tells me failure to sync. I restarted it, but Bit Torrent has got to be the faster more stable solution. So I think I can remote to a faster stable connection download it then seed a torrent I can download. Be nice to get it on http://linuxtracker.org/ and get more seeds too.

Thankfully, I had some offline FreeCAD tutorials to work on during all this so, progress. The nature of the work benches is an interesting way to break workflow down, but some of the inconsistencies and copy limitations are making it hard for me to see the optimal order of operations for mirroring symmetry and reducing repetition. Many tutorials how a lot of repetition, which is good for learning, but not optimal for productivity. However, the OSE distribution methods, small simple parts in libraries and assembly make these issues less critical. Also, the software is still beta. So next I think I'll do something with a focus on the assembly work benches.

While the FreeCAD Developer Test doesn't look too hard, I think the broader point is to delve deeper into all the OSE Linux software so anyone can complete a series of. I was thinking about the software development side of things more lately, but the open source software community is already huge and often willing to mobilize on good projects quickly so perhaps an if we build it they will come approach might work out. It is currently the open source hardware side of things that needs more development and growth. While it has to be a team effort there are many good existing software dev teams capable of joining if they see a capacity for steady hardware production for them to write software for.