Suggestions for Possible Features...

[wwderw]

I first want to say I'm glad that I found this project. I've been a freelance digitizing for many years (23.5 yrs now), running Wilcom mainly, and it was hard to find open source programs that had a good feature set.

I've tried Thred (limited due to the time period it was written, although being able to work with the CSV version of the THR file helped improve that to some degree), embroiderymodder was too alpha, so I've always had to stick with Wilcom and VM Windows on my Linux boxes.

I've tried the extension for a few days and I'm liking what I've seen. Just wanted to see if a few things were possible. I wish I could code to try to do myself, but I'm an idiot when it comes to coding.

1. With regard to the python scripts, is there anyway to install those scripts without having to install everything else to grab those dependencies? I try to run as lean of an install as I can and since I have no use for pip and the like elsewhere since I'm not a developer and just uninstalling those other programs doesn't give back a lot of space.

Is there anyway to bundle them in within the extension itself? As a non coder, I don't know if that's possible or not. Just making the extension as portable as possible would be my way of thinking.

With regard to the function of the extension:

2. Is there anyway that we can set it up to where we can insert functions in specific objects. Trim, stop, jump, drop sequin (I do have a sequin attachment on one of my machines) etc?

3. Is there any way we can have the option of making a vector fill object either a fill stitch or satin stitch. Some of the more complex turning satin stitches won't easily be done with just using strokes to tell it to do a satin stitch. Strokes wouldn't have the same ability as a fill stitch, but it also has to be able to handle multiple angles, not just the static one angle (hence being called turning satin stitches).

4. Also is there way to have an option for both RGB color values and a generic color name as well for a stitch list? I'm thinking more as a freelance digitizing and those that use DST files, I have to send a stitch chart out to my customers.

Possible or is my coding ignorance really showing itself here?

Thanks again for a wonderful extension and glad to see that it's still active.

[lexelby]

Hey there! Welcome, and thanks for the suggestions. I'm really excited to have someone who's actually been doing this for a long time and knows what they're actually doing. I'm completely self-taught and have never done digitizing commercially, and while I've learned a few things about underlay and satin and the like, I'd love to fill in the gaps in my knowledge.

For your question 1, there's already work underway by @Moini in #5 to try to make installation easier. I hear you about dependencies taking up space, but packaging the dependencies into one big blob isn't going to free up space. Perhaps Moini's script might install only the bare minimum you need to run my extension, though?

Question 2: Absolutely. The thing is, I don't have a machine that can do TRIM, much less sequins. I really long for a machine that can cut the thread and continue on, but my little brother SE400 just isn't capable of that. I could try to add the capability to put in additional instructions, but honestly, without a machine to test them on, I'm not sure how I can effectively ensure that it works.

Question 3: Here I'm running into the limitations of my knowledge of machine embroidery terminology. I think you're asking this: can I do a proper satin column, that is, zig-zag between two non-parallel edges, rather than a simple zig-zag along a path? The answer to that question is yes! It's explained in the main README document in this repo, although the lack of pictures makes it unclear. My extension easily supports satins like this:

I modeled my satin column and its underlay settings off of Mr Xstitch's excellent description.

If that's not what you meant, can you give an example picture of what you're talking about?

Question 4: Generic color names would be a bit trickier, but that's really just a matter of the underlying software library I use for writing file formats like DST, PES, etc. I use libembroidery which is part of the Embroidermodder project you mentioned. Embroidermodder never really got fully fleshed out, but at the core of it is an excellent library of code that knows how to read and write a TON of machine embroidery formats! I'd have to look into the DST format writer to see how it can handle colors.

Bottom line: yes, what you're asking for is possible. I don't pay much attention to color because my Brother machine gives me two options: display the color or display the current stitch number, and I chose the latter. :) I suppose I'd feel differently if I were trying to publicize my designs.

Thanks again for your suggestions here, and I really hope we can make this extension into something you can use for your work!

[wwderw]

Hey there! Welcome, and thanks for the suggestions. I'm really excited to have someone who's actually been doing this for a long time and knows what they're actually doing. I'm completely self-taught and have never done digitizing commercially, and while I've learned a few things about underlay and satin and the like, I'd love to fill in the gaps in my knowledge.

If you want to see some examples of my stuff you can go here.

If there is anything that I can help you with, please don't hesitate to let me know. I love talking about this stuff.

For your question 1, there's already work underway by @Moini in #5 to try to make installation easier. I hear you about dependencies taking up space, but packaging the dependencies into one big blob isn't going to free up space. Perhaps Moini's script might install only the bare minimum you need to run my extension, though?

I'm sorry, I think I misspoke. It's not so much the size of dependencies, it's trying to keep having to install the other stuff every time. My usually work flow when compiling and deploying is I usually use either a dedicated desktop just for compiling from source or a VM on one desktop and then compile and build and from there, take what has been built and move that on over to other computers.

I'm not all that concerned about the dependences being X amount, but rather after I get them on one computer don't have to go through the process on another, just be able to move them after everything is done to the next computer. Make it easier for deploying across many computers without having to say install pip to install the dependences etc.

Kinda like after you build libembroidery-convert, you can take that executable file to any computer (that shares the same arch) and use it. I don't have to recompile, build etc again. Does that make sense?

Question 2: Absolutely. The thing is, I don't have a machine that can do TRIM, much less sequins. I really long for a machine that can cut the thread and continue on, but my little brother SE400 just isn't capable of that. I could try to add the capability to put in additional instructions, but honestly, without a machine to test them on, I'm not sure how I can effectively ensure that it works.

I wouldn't mind running tests on my machines. I've got one PR machine that usually is just used for 1 offs that I could run tests on to see how it handles the functions. If we managed to get sequin functionality, I would have to use the big machine for that as it's the only one with sequins.

Question 3: Here I'm running into the limitations of my knowledge of machine embroidery terminology. I think you're asking this: can I do a proper satin column, that is, zig-zag between two non-parallel edges, rather than a simple zig-zag along a path? The answer to that question is yes! It's explained in the main README document in this repo, although the lack of pictures makes it unclear. My extension easily supports satins like this:

I do believe that's it. I'll have to take a look at the readme file again and see how to do it and experiment with it for a few days, but I do believe that'll get me covered.

Along the lines of that, what about being able to assign a specific spot that you want it to end at? Let's say I wanted it to end in the middle of the object at the left side?

Question 4: Generic color names would be a bit trickier, but that's really just a matter of the underlying software library I use for writing file formats like DST, PES, etc. I use libembroidery which is part of the Embroidermodder project you mentioned. Embroidermodder never really got fully fleshed out, but at the core of it is an excellent library of code that knows how to read and write a TON of machine embroidery formats! I'd have to look into the DST format writer to see how it can handle colors.

I use the libembroidery-convert to actually take Thred's THR format, make it a CSV and manually add trims to the files there. I know you can prompt a color, just by entering COLOR in place of STITCH, but it doesn't assign the actual number to the color and I haven't been able to find the syntax with the conversion program that it will take and pass it to converting it back to a raw stitch file. I do have an issue with that project to find out how it's support, the syntax etc.

Thanks again for your suggestions here, and I really hope we can make this extension into something you can use for your work!

I hope so too. Like I said, I cut my teeth with Wilcom and I love the program, but since I've moved to Linux, about the only thing that I have to use a VM for is that program. Thred will run in WINE, but it has it's limitations with my needs and it's no longer active. EmbroideryModder was my first hope, but it never got out of alpha.

I discovered this project a few days ago and was really impressed and I'm glad that it seems to be thriving. Linux seems like it has a lot of alternatives except with the embroidery software, so I'm really rooting for this.

[lexelby]

I'm sorry, I think I misspoke. It's not so much the size of dependencies, it's trying to keep having to install the other stuff every time. My usually work flow when compiling and deploying is I usually use either a dedicated desktop just for compiling from source or a VM on one desktop and then compile and build and from there, take what has been built and move that on over to other computers.

I'm not all that concerned about the dependences being X amount, but rather after I get them on one computer don't have to go through the process on another, just be able to move them after everything is done to the next computer. Make it easier for deploying across many computers without having to say install pip to install the dependences etc.

Kinda like after you build libembroidery-convert, you can take that executable file to any computer (that shares the same arch) and use it. I don't have to recompile, build etc again. Does that make sense?

Ah, I understand now. It should be possible to install all the dependencies in a self-contained directory using virtualenv. Then you can just copy the whole directory over from one machine to the next. That'd solve the Python side. For libembroidery-convert, yes, you could compile it statically and copy it over. Ideally, I'd like to make libembroidery into a debian package, and create an additional libembroidery-tools package to contain libembroidery-convert. I've just never gotten around to it.

I wouldn't mind running tests on my machines. I've got one PR machine that usually is just used for 1 offs that I could run tests on to see how it handles the functions. If we managed to get sequin functionality, I would have to use the big machine for that as it's the only one with sequins.

That'd be great! I'm really not sure when I'll get to this, as my availability to work on this project can be really sporadic.

I'm not sure how this kind of command thing would work in Inkscape. Any ideas? I could add it as a parameter in the Embroidery Params dialog, but then it wouldn't be visually obvious that a trim/sequin is happening...

I do believe that's it. I'll have to take a look at the readme file again and see how to do it and experiment with it for a few days, but I do believe that'll get me covered.

Along the lines of that, what about being able to assign a specific spot that you want it to end at? Let's say I wanted it to end in the middle of the object at the left side?

With this extension as it stands, you'd break that satin in half and add a line of running stitch between them. I did that for some of the letters in my mini font before I got bored of it and realized it was actually easier to trim longer jumps. :)

As to ending on the left or right, I control that by changing how I construct the satin in Inkscape. You define a satin column by creating a path with two subpaths, e.g. the left and right sides of the S in the picture above. If you have two paths, you can join them together into a single path by selecting both using Inkscape's Path -> Combine command. If you select the left first, then the right, and do Combine, then my extension will start the satin on the left and end on the right.

It's not exactly intuitive, but I have to work with Inkscape's interface, since extensions can't modify the interface at all.

Now, I'm getting from your question that maybe Wilcom and other programs allow you to give them a starting and ending point for a satin. How does that work? Does it automatically fill an entire shape with satin, figuring everything out itself? My code definitely doesn't do that -- I haven't figured out what that kind of feature would look like yet, much less how to code it. Would you be willing to make a quick screen recording showing how that looks in wilcom?

I hope so too. Like I said, I cut my teeth with Wilcom and I love the program, but since I've moved to Linux, about the only thing that I have to use a VM for is that program. Thred will run in WINE, but it has it's limitations with my needs and it's no longer active. EmbroideryModder was my first hope, but it never got out of alpha.

I discovered this project a few days ago and was really impressed and I'm glad that it seems to be thriving. Linux seems like it has a lot of alternatives except with the embroidery software, so I'm really rooting for this.

Yup, I went through a similar search for Linux embroidery software about 2 years ago, with the added constraint that I really didn't want to pay hundreds or thousands (!) of dollars for embroidery software to start a hobby I might not even enjoy! And then I discovered a much earlier version of this extension and started hacking features into it, and it turns out I enjoy coding for embroidery just as much as actually designing stuff. :)

Some of the designs I've made are here. If you have a moment, I'd really appreciate critique from an experienced eye.

[X3msnake]

Hello i'm interested in the cut feature too.

I have acess to a happy embroidery machine with cut feature. And I am trying to define a workflow for the machine so people can learn and do embroidery.

This project is the only one active these days and i'm very happy to test the cut feature.

I am self taught also but have design background and some notions of programming

Normally on the embroidery programs you have a control box with arrows that allow you to navigate the jump stitches. There you can choose to add a cut.

Haven't tried to read the code you have written yet but I guess if we have a array of all the stitches and add a jump tag that corresponds to the stitch index nr to a new array we can check and push the modifier in the export loop.

No dia segunda-feira, 4 de dezembro de 2017, Lex Neva < notifications@github.com> escreveu:

I'm sorry, I think I misspoke. It's not so much the size of dependencies, it's trying to keep having to install the other stuff every time. My usually work flow when compiling and deploying is I usually use either a dedicated desktop just for compiling from source or a VM on one desktop and then compile and build and from there, take what has been built and move that on over to other computers.

I'm not all that concerned about the dependences being X amount, but rather after I get them on one computer don't have to go through the process on another, just be able to move them after everything is done to the next computer. Make it easier for deploying across many computers without having to say install pip to install the dependences etc.

Kinda like after you build libembroidery-convert, you can take that executable file to any computer (that shares the same arch) and use it. I don't have to recompile, build etc again. Does that make sense?

Ah, I understand now. It should be possible to install all the dependencies in a self-contained directory using virtualenv. Then you can just copy the whole directory over from one machine to the next. That'd solve the Python side. For libembroidery-convert, yes, you could compile it statically and copy it over. Ideally, I'd like to make libembroidery into a debian package, and create an additional libembroidery-tools package to contain libembroidery-convert. I've just never gotten around to it.

I wouldn't mind running tests on my machines. I've got one PR machine that usually is just used for 1 offs that I could run tests on to see how it handles the functions. If we managed to get sequin functionality, I would have to use the big machine for that as it's the only one with sequins.

That'd be great! I'm really not sure when I'll get to this, as my availability to work on this project can be really sporadic.

I'm not sure how this kind of command thing would work in Inkscape. Any ideas? I could add it as a parameter in the Embroidery Params dialog, but then it wouldn't be visually obvious that a trim/sequin is happening...

I do believe that's it. I'll have to take a look at the readme file again and see how to do it and experiment with it for a few days, but I do believe that'll get me covered.

Along the lines of that, what about being able to assign a specific spot that you want it to end at? Let's say I wanted it to end in the middle of the object at the left side?

With this extension as it stands, you'd break that satin in half and add a line of running stitch between them. I did that for some of the letters in my mini font https://github.com/lexelby/embroidery-fonts before I got bored of it and realized it was actually easier to trim longer jumps. :)

As to ending on the left or right, I control that by changing how I construct the satin in Inkscape. You define a satin column by creating a path with two subpaths, e.g. the left and right sides of the S in the picture above. If you have two paths, you can join them together into a single path by selecting both using Inkscape's Path -> Combine command. If you select the left first, then the right, and do Combine, then my extension will start the satin on the left and end on the right.

It's not exactly intuitive, but I have to work with Inkscape's interface, since extensions can't modify the interface at all.

Now, I'm getting from your question that maybe Wilcom and other programs allow you to give them a starting and ending point for a satin. How does that work? Does it automatically fill an entire shape with satin, figuring everything out itself? My code definitely doesn't do that -- I haven't figured out what that kind of feature would look like yet, much less how to code it. Would you be willing to make a quick screen recording showing how that looks in wilcom?

I hope so too. Like I said, I cut my teeth with Wilcom and I love the program, but since I've moved to Linux, about the only thing that I have to use a VM for is that program. Thred will run in WINE, but it has it's limitations with my needs and it's no longer active. EmbroideryModder was my first hope, but it never got out of alpha.

I discovered this project a few days ago and was really impressed and I'm glad that it seems to be thriving. Linux seems like it has a lot of alternatives except with the embroidery software, so I'm really rooting for this.

Yup, I went through a similar search for Linux embroidery software about 2 years ago, with the added constraint that I really didn't want to pay hundreds or thousands (!) of dollars for embroidery software to start a hobby I might not even enjoy! And then I discovered a much earlier version of this extension and started hacking features into it, and it turns out I enjoy coding for embroidery just as much as actually designing stuff. :)

Some of the designs I've made are here https://github.com/lexelby/inkscape-embroidery/tree/master/images/patches. If you have a moment, I'd really appreciate critique from an experienced eye.

— You are receiving this because you are subscribed to this thread. Reply to this email directly, view it on GitHub https://github.com/lexelby/inkscape-embroidery/issues/6#issuecomment-348837426, or mute the thread https://github.com/notifications/unsubscribe-auth/AKke-iYxs5v_oDTXASZcI8MnL_Mal5oqks5s80kEgaJpZM4QzjO3 .

-- Com os melhores cumprimentos, Vinicius Silva

[wwderw]

I'm not sure how this kind of command thing would work in Inkscape. Any ideas? I could add it as a parameter in the Embroidery Params dialog, but then it wouldn't be visually obvious that a trim/sequin is happening...

I would just have individual functions as bottoms to be applied and then one would have to look at the XML editor to see if they carried through or not.

This is pretty much how it's handled in programs like Wilcom, you have to look at the stitch list module to see what functions were used/when applied (akin to looking at the CSV format of a file).

With this extension as it stands, you'd break that satin in half and add a line of running stitch between them. I did that for some of the letters in my mini font before I got bored of it and realized it was actually easier to trim longer jumps. :)

That's pretty much what I would have to do in Thred, is break the object down if I wanted specific spots to end. Now, sometimes it would generate the end point on the opposite side, in which case, I would just use the manual stitch and do one stitch back over and then continue on from there.

Now, I'm getting from your question that maybe Wilcom and other programs allow you to give them a starting and ending point for a satin. How does that work? Does it automatically fill an entire shape with satin, figuring everything out itself?

A little bit of both. Some tools will initially have start/end points dictated, some will allow you to create them on the fly. However, depending on what version of the software you are running, it will allow you to go into "reshape mode" and move those start/end points on the fly.

Normally on the embroidery programs you have a control box with arrows that allow you to navigate the jump stitches. There you can choose to add a cut.

Typically, so many jump stitches in a row (for DST that's usually "3", for PES, I think it's "4" or "5") will invoke a trim. That's how I get away with it in csv files generated from Thred that don't have jump commands period (due to home machines that didn't have that functionality at the time).

[wwderw]

Ah, I understand now. It should be possible to install all the dependencies in a self-contained directory using virtualenv. Then you can just copy the whole directory over from one machine to the next. That'd solve the Python side.

I just got around to trying this, that did not work. I still got the same error as if I didn't have those files.

Just in case I did something wrong, here were my steps.

Download virtualenv, created new directory and activated it

Used pip to download requirements.txt into active folder

Made it relocatable

Moved to another computer

Then I tried using the extension. Is there something that I need to do at the new computer with that folder with scripts? Seems to me like there should be, but nothing in the instructions on their site mentioned such.

Thanks.

[Moini]

What is the OS on your computers? What is the error message you get?

[wwderw]

What is the OS on your computers? What is the error message you get?

Mint 18

It's the standard error message that one would get if the dependencies weren't on the computer at all. It's like it's not reading the folder that has the dependencies in it.

It seems to me like the system isn't "reading" that it is there. I put it in the exact place that I had installed it on the original computer.

Now, I must say that I'm using Inkscape as an AppImage and not directly installed on the rig itself. I don't know if that would have any bearing, but I thought I would mention it.

[Moini]

You could try if the script works for you (see README), or look into the script how it is done. It has been developed on Mint 18.2 :) There's no building needed, except for embroidermodder, if you want to install that additionally.

[wwderw]

I'm not a programmer by any stretch of the imagination, so if I'm really messed up in my thinking/hopes, I apologize for that.

You could try if the script works for you (see README), or look into the script how it is done. It has been developed on Mint 18.2 :)

It won't remove the AppImage as it isn't a snap package, but that's easy enough for me to delete on my own.

Now, with regard to this last line of the script:

cp *.py *.inx $HOME/.config/inkscape/extensions

Will this move everything that is needed, even the dependencies into the extension folder or just the extension itself?

The main thing is to be as portable as it can be. Right now, I'm just moving the build folder of WINE and running Thred within that as it's easier for me to move that among different computers.

There's no building needed, except for embroidermodder, if you want to install that additionally.

I don't mind building if it's just building on one computer and easily able to deploy it on others. Like I do with Wine, but also with libembroidery-convert program as well. Just one little file that moves from one computer to the next.

[Moini]

It's just going to move the extension there. But isn't it possible to deploy the script?

Or do the other machines not have access to the web? Then you can just copy the deb packages and the zip file, and then install / unpack them, using dpkg, unzip and cp.

[wwderw]

Or do the other machines not have access to the web?

Isolated production machines (embroidery production, access to client files etc). Another reason why I like to run thin installs.

I don't necessarily understand why creating that individual folder with the python dependencies didn't work.

Is there some command or something along those lines that I need to do tell the system when it's getting calls to these dependencies to look at this folder (I hate not knowing all this "under the hood" stuff, my mind just doesn't go there though).

[Moini]

Can you list a bit more explicitly what exactly you copied where and which packages have been installed on your offline machines?

No, there is no extra command. Inkscape will find it, when it's in the right place. Look into the extensions log file (in /home/.config/inkscape) to see any error messages.

Is package installation an option on those machines?

[wwderw]

Can you list a bit more explicitly what exactly you copied where and which packages have been installed on your offline machines?

I used Virtualenv to put all the dependencies in the requirements.txt file into their own file (it was active) on my home directory.

I made that directory relocatable in order to move it to another computer.

Is package installation an option on those machines?

I use AppImage, so I can bundle everything together, including extensions (I have a couple, this one and InkCut, plus some customer color palettes) and this just foregos the need to all that relocations after a package install and setting up everything all over again, when it's done once and that's it.

Nothing in the extension error log that referenced the extension issue. The extension just keeps on telling me that they additional python modules aren't loaded, but they are in the folder on the home directly. Is there anyway to manually point those extensions to look in that directory? It would be static on my end anyway, I would always put the same folder tree in the home directory, so doing that once wouldn't be an issue.

[Moini]

So, to clarify: you've only got a virtualenv on your computer with the python dependencies? I'm not sure if Inkscape is actually going to use that, even if you have activated the virtualenv and are starting Inkscape from there. I think it might be going to be opening a new shell that does not inherit the environment - you could make any of the working extensions print out the location of the actually used python files to confirm/negate this.

You've made the AppImage of Inkscape yourself? Or where/how do you add the extensions?

Is imaging an option? I.e. will all the computers you have there be provisioned with the exactly same set of software?

I'd probably not use a python virtualenv, but would just collect the deb packages, the python eggs and the extension py and inx files, zip them up, and bundle them with a script that installs them.

[lexelby]

I would just have individual functions as bottoms to be applied and then one would have to look at the XML editor to see if they carried through or not.

This is pretty much how it's handled in programs like Wilcom, you have to look at the stitch list module to see what functions were used/when applied (akin to looking at the CSV format of a file).

Okay, that makes sense. I could add a checkbox to the Params dialog along the lines of "Trim thread after". The actual implementation of the TRIM instruction in the code is trivial.

[lexelby]

That's pretty much what I would have to do in Thred, is break the object down if I wanted specific spots to end. Now, sometimes it would generate the end point on the opposite side, in which case, I would just use the manual stitch and do one stitch back over and then continue on from there.

That would work with this extension too, if you didn't want to worry about which order you linked the two paths together in.

A little bit of both. Some tools will initially have start/end points dictated, some will allow you to create them on the fly. However, depending on what version of the software you are running, it will allow you to go into "reshape mode" and move those start/end points on the fly.

Gotcha. "Reshape mode" is the kind of thing that'd be pretty tricky since inkscape-embroidery can't interact with the user directly. It sounds like maybe having to shape it yourself isn't all that bad, though. Let's put a bookmark in this and come back to it later if it turns out to be too annoying of a user experience as it is.

[lexelby]

Thanks for jumping in to help debug, Moini!

@wwderw I know what's going wrong with the virtualenv. The problem is that Inkscape runs the python interpreter itself so the PYTHONPATH doesn't point to the virtualenv. It may be enough to simply activate the virtualenv in a shell window and then run inkscape from the command-line so that it has the proper PYTHONPATH.

A brief look through the inkscape source code suggests to me that this should indeed solve your problem. Could you give that a try and report back? If it doesn't help I have a few other tricks up my sleeve.

[wwderw]

I'd probably not use a python virtualenv, but would just collect the deb packages, the python eggs and the extension py and inx files, zip them up, and bundle them with a script that installs them.

The script itself is something that's beyond me.

I did try to collect the deb files, but networkx and wx are actual compressed files for compiling, so I stopped in that effort there.

Okay, that makes sense. I could add a checkbox to the Params dialog along the lines of "Trim thread after". The actual implementation of the TRIM instruction in the code is trivial.

I think that sounds like a good plan for all the different types of functions, to have a check box for after, before etc. I would suggest the actual functions as a DST file would implement them, since that would apply to more machines, even alot of home machines can read that format.

The only concern that I would have is does libembroidery understand those codes beyond what it prefaces at the beginning of it's CSV file? I know color changes in THRED show up as UNKNOWN when converted to CSV. Now the simple change would be to go from UNKNOWN to COLOR, but I'm hoping that (especially for THR files) can actually do something like COLOR (8) as that syntax is understood by stitch files (especially when going to PES, Jef, PCS etc) that do preserve color functions. This isn't a dire need for single needles, but for multi-needles it's helpful.

A brief look through the inkscape source code suggests to me that this should indeed solve your problem. Could you give that a try and report back? If it doesn't help I have a few other tricks up my sleeve.

I'll give that shot when I'm able and report back.

[wwderw]

A brief look through the inkscape source code suggests to me that this should indeed solve your problem. Could you give that a try and report back? If it doesn't help I have a few other tricks up my sleeve.

Same errors. I even moved the appimage to the bin file that I would activate to see if for some odd reason that might help.

It may not be able to be made "portable". Those dependencies anyway.

[lexelby]

I wonder if we can be sneaky with the virtualenv. When you create a virtualenv, along with bin/activate, you also get bin/activate_this.py. The instructions say that you can add this to your script:

execfile("path/to/your/virtualenv/bin/activate_this.py")

to activate the virtualenv in an existing, running script. This is useful in cases where you can't change the python interpreter, which is exactly the position we find ourselves in here.

That means we could create pre-built ZIP files for Linux, OS X, and Windows, each containing its own virtualenv with all of the necessary dependencies. The zip file would have a structure like this:

embroider.py
embroider.inx
embroider_params.py
embroider_params.inx
[etc]
embroider_virtualenv/

Our installation instructions would be to unzip the file directly into Inkscape's extension directory. That way, the scripts could easily find their virtualenv with all of the dependencies.

In fact, we could even package inkstitch as a python module itself. Then the scripts at the top level (embroider.py, etc) would simply activate the virtualenv, import inkstitch, and run it.

Thoughts?

[wwderw]

If we are able to do that, I would say that would be a good way to do it.

That way it would be really easy to put that in to an AppImage of Inkscape as well, into it's extension directory. Making deployment among Linux boxes much, much easier.

For Windows, it would be easy enough. I'm not allowed near the "ball and chains" MacBook, she is too afraid that it will be booting to Linux, so I can't experiment with that one.

[lexelby]

I'll give this idea a try when I get a chance. The especially interesting thing here is wxPython. If we install it with Pip (rather than using the ubuntu packages) then I think there won't be any need for the user to install wxgtk separately. That'd be a huge win.

[probonopd]

You can bundle very complex Python based applications as AppImages that run on most Linux distributions. The key is to bundle Python and all the modules that are needed for the application to run. Ultimaker Cura, for example, is written in PyQT and is distributed as an AppImage for Linux: https://ultimaker.com/en/products/ultimaker-cura-software

Let me know if you are running into issues.

[lexelby]

I'm going to close this, since we've implemented a good chunk of it. I'll add a new feature request for color naming.