Consider expanding or dropping (more likely) certain characteristics for which data is unreliable or null.

For example "leaf" column contains majority null values. First check that this isn't a code issue, and if not, consider the usefulness of the data. Could prove to be valuable information even if just 10% of entries have it. In which case we could ask for user feedback to fill data.

Also consider possible categories for useful future data. The more (reliable) data the better!

As we continue to scrape from other sources like permapeople, pomiferous, etc..., we'll find many characteristics of one database are undefined in the other. Discuss and evaluate how to combine databases.

in UI and data, and in dimensions (sizes US-EU)

EN would be reference, FR, DE and ES good candidates for translations

Customize grid dimensions to design size (user input) and row spacing.

May be helpful for ticking grid lines.

Also possibly helpful for calculating overlap % between plants.

to test and help discuss relevance+exact content of data.
help define mapping from external sources, via python scrapers or google refine scripts or other APIs...
likely goolesheet.

Somewhere the genus and species are being split with a tab \t and every plant results in an error.
I've tried splitting genus and species seperately and then stripping with .strip() but can't figure it out.

Running test_plants.txt returns this:

Errors: ['Bunium\tbulbocastanum ', 'Calamintha\tnepeta ', 'Carex\tpensylvanica ', 'Carya\tillinoinensis ', 'Gymnocarpium\tdryopteris ', 'Halesia\ttetraptera ', 'Juglans\tnigra ', 'Juglans\tregia ', 'Malus\tpumila ', 'Malva\talcea ']

Many of the "Width" entries in pfaf are null, and it is essential to create the design.

So we need to create a reasonably accurate width estimator (Finally get to use some data analysis skills) :)
My initial idea is to create a height-width ratio for all plants. I assume trees have a larger ratio (more heigth to width) than shrubs so we could return a different ratio for different plants. Or look at the heigh-width ratio of members in the taxonomic family?

Would love to be able to collect+merge existing sources and provide a means for a community of users to curate/improve it.

That collaborative evolution process does not need to be in the app: this 'reference' data could live somewhere online (along with translations, tips, anecdotes, pictures, seeds/plants local providers coordinates, or feedback) and be exported to the app on new release or manual update.

Sources already considered for import/inspiration:

• https://pfaf.org
• https://permapeople.org plants database (and its many filters) **This one I still need to scrape

Sources for specific fruit cultivars:

• https://orangepippin.com
• https://pomiferous.com (Even more thorough than orangepippin
  Others to include pears, stones fruits, etc...

Sources maybe interesting but not yet scrapped:

• Edible Forest plants set, hundreds of edible or useful plants, readily re-encoded at https://docs.google.com/spreadsheets/d/1_PgxV4pxlNTa0Ep9TDY_57oJfVtrofmu/edit#gid=1609043836$-
• roots systems : https://images.wur.nl/digital/collection/coll13/search/searchterm/wurzelatlas%20:%20mitteleuropaeischer%20ackerunkraeuter%20und%20kulturpflanzen/field/subcol/mode/all/conn/and/

Sources unlikely to be integrated in anyway, but that may be interesting for permadesigners, for info:

• http://www.gartendatenbank.de/wiki/ along with good gardening tips
• https://www.plantopedia.de/pflanzenbestimmung/ and their plant recognition indexes based on observable characteristics
• https://alveoles.fr/fiches-plantes for their beautiful paper cards. Their online database provides good info, though no sexy looks: https://plantes.universite-alveoles.fr and lacking some filters to find good companion plants.

We'll need svg images for different tree / shrub / plant forms. Obviously not all plants will have a unique image but a certain number of categories will. Think about the "habit" column in this EFG database:

We have to decide whether to use a minimalistic circle or an svg with a bit more detail to include a wavy outline, some branches, etc. I would like to use more than a simple circle but without adding weight to our program with complex svg's.

And some variation would be nice so that not all trees look the same. We could rotate the same design (unless it includes shadow) but it shouldn't be hard to add some variety to our options.

Another idea would be to include an icon with each tree to differentiate it. For example a chesntut tree would have a chestnut icon, an apple tree an apple, an oak and acorn, and so on. We could include icons for nitrogen fixers or pollinizers too. I don't know that I like this idea but it is something to consider!

what would show on first page ?

• user name to "sign" the designs ? < multiple accounts on single device ?
• recent designs to open ? action to start new design ?
• 2-3 "random" plants for discovery ?
• designs by other users?
• technical tool infos (version, update notifications,...) ?
• settings ?
• help/instructions/guided tour ?
• links to stocks/providers/other infos ?

https://kivy-garden.github.io/mapview/mapview.html

Seems to allow displaying an interactive map as background, openstreetmap based by default.

Might be cool to be able to import+export from/to GIS software.

Check if

• other tiles engines might be available, eg for HD satellite view, eg test with Luxembourg opendata geographic data EU Inspire
• ease to move other interactive widgets on top (markers?)
• able to add grid on top (automatically or using a JSON layer?)
• able to show level lines
• light enough on hardware resources (important)
• usable offline (caching...)
• export to GIS (kml, geoshapes?)

I put it with target 1.3 for now, but if you feel it's worth looking at earlier just change the milestone :⁠^⁠)

Create mechanisms for users to edit their own database. And down the line allow for these edits to be added into the main database, either as user comments or entries themselves. Perhaps a voting system?

No filters for now. Just create search function searching by name. Have results (scrollable if applicable) drop down from search bar and select one by clicking. Or hit enter and be able to select from all results, filtering where necessary (see next step). On mobile I think 'Accordion' is the class for this, and maybe on desktop too.

Integrate the search with the UI in design directory.

VERY useful, thanks for sharing !
I'll keep following for the next steps.

my favorite (online design) tool so far is https://permapeople.org/berteh/appel-tree-pattern/

I like there :

• easy grid-based design, makes alignment quick
• dragn'drop move
• full text search to find plant on any characteristic/name/language
• remembers last added plant, so adding more of it goes quick

I don't like

• that i don't see the space the mature plant will occupy (eg projected circle), a major design factor for me
• lack of free move to place the plants in roundish layouts. Nature is never square.

I miss there :

• export of the design with amounts and locations (only export so far is CSV list of latin names)

add some timelines view in a design export that combines

• all their flowering periods (directly with plants colours ?) (on their respective heights ?)
• all harvest periods
• recommended seeding/planting period

Create 2-3 filters/facets to understand the process. Using the 'working data' from design_data_class these could be pH, habitat, and plant size.

story : i have "X" plants of type"W" and "Z", please generate a design from that

with minimal infos needed to be able to display the design in editor, and intuitive access methods, likely stored in a pandas DataFrame (but pick whichever internal structure you feel more relevant).

some thoughts relating to data storage

• be able to use reference plants data (eg from scrapers you already made) loaded by the app.
• all permadesigner to define/refine his own pet plants, local varieties or local calendars that vary a bit from the reference. They should not replace but rather 'superseed' the reference data (like an 'inheritance' mechanism). That should be usable/copyable/refinable in the app in some fast way, maybe with an ability to 'suggest for curation' to the reference dataset, either automatically (as a 'give back to community' mechanism) or manually.

About the data storage technology : how big do you expect it will become? How many plants and designs would be stored in the user's app?

If the sum is less than 2000 then the technology doesn't matter much I think.

Do you need instant online sync amongst all users? I think an update every once in a while is enough and thus wouldn't go for online CRUD like restdb, but I'd rather stick to offline & local file with some import/update mechanism.

In case of local files : SQLite is easy enough but really low level, opt. use sqlalchemy for ORM or MVC or later backend change... but I wouldn't invest too much time there in now. Parquet file is a good alternative to CSV, that natively loads to pandas data frame.

Add a layer on top of background and below plants to draw structure elements (paths, ponds, walls).

Could be a layer to import, scale and rotate various transparent images, or drawing/scribbling layer.

Hi Nigel, I am a developer interested in permaculture and I have found your project very interesting. I had in mind something similar to what you are doing. I have started with an application in Python for companion plants, you can see it here:
https://github.com/RPDaniels/CPlant

My idea was to built a sw suite for permaculture design, and I am currently scanning the web to check the current state of the art of permaculture projects for not to reinvent the wheel.
It would be great if we can gather all the functionalities in a single application.

I will have a look at your current development
Cheers from Spain!

Hi nigel.
The following static paths may be problematic to deploy/use elsewhere:

$ grep -Ir  "/home/nigel" *
design/libs/screens/db_screen/__init__.py:data = pd.read_csv('/home/nigel/Code/permaculture/scrapers/pfaf/all_plants.csv')
design/libs/screens/db_screen/__init__.py:        db_data = pd.read_csv('/home/nigel/Code/permaculture/scrapers/pfaf/all_plants.csv')
design/env/bin/pip3.10:#!/home/nigel/code/jwnigel/permaculture/design/env/bin/python
design/env/bin/activate.fish:set -gx VIRTUAL_ENV "/home/nigel/code/jwnigel/permaculture/design/env"
design/env/bin/activate:VIRTUAL_ENV="/home/nigel/code/jwnigel/permaculture/design/env"
design/env/bin/pip:#!/home/nigel/code/jwnigel/permaculture/design/env/bin/python
design/env/bin/activate.csh:setenv VIRTUAL_ENV "/home/nigel/code/jwnigel/permaculture/design/env"
design/env/bin/pip3:#!/home/nigel/code/jwnigel/permaculture/design/env/bin/python
design/env/pyvenv.cfg:home = /home/nigel/.pyenv/versions/lewagon/bin#40 

After the initial design is done, I would love to be able to export a list/table of the selected plants with their respective:

• amount in that particular design
• ecological function
• human possible uses
• strate/layer
• colour shade

• laptop view : https://balsamiq.cloud/szax2i8/pfszbl1/r627C
• tablet view
• smartphone view : https://balsamiq.cloud/szax2i8/pfszbl1/r74A1

now available at https://github.com/users/jwnigel/projects/1/views/3?filterQuery=-status%3ADone&sortedBy%5Bdirection%5D=asc&sortedBy%5BcolumnId%5D=Milestone&visibleFields=%5B%22Title%22%2C%22Assignees%22%2C%22Status%22%2C%22Milestone%22%5D

• Still getting this error sometime although search term works fine in browser: invalid literal for int() with base 10: ' ' ***Hardiness zone fixed this issue. I think it was zones >= 10 throwing error. ***Using float() instead of int() also.

• Height is returning the string 'to' sometimes. I noticed it in Onion and Rosemary.

• In Habitat (maybe others), remove pfaf references in square brackets [] Example: 'Rich fertile woods and hillsides[43, 82] in deep well-drained soils[229].' (Although maybe this is necessary for liscensing?

• Clean up output in general: Sometimes strings like "perennial" are returned in all caps. Make things consistent.

• Use regex to grab hardiness zones. Current [-1] indexing doesn't work when zones >= 10

• Add pollinators to scraper

• When scraping varieties like Acer saccharum var nigra, the search string should be 'Acer+saccharum+nigra', without 'var'

• #32

export to some

• generated "artist-like" sketch (from a given direction) that would compose all plants with their respective heights and widths at maturity.
• it could include the roots systems types likewise... for more educational purposes

• After dragging and dropping trees to the screen, upon resizing the window the trees move dynamically with the RelativeLayout. However, they do not move correctly with the background image! In order to keep the image proportions, padding is added to the top/bottom or sides of the image as the window is resized. I think one solution to this would be to crop the image instead of padding it.
• When I click on the tree icon (IconButton class) it creates a DraggableImage in the MapLayout directly over the button's location. However you have to click on this again to drag it. Need to set it's default drag state to True somehow. I already tried using on_motion property.
• Resize the tree (DraggableImage) drag size. Right now the size is set to the size of tree IconButton, which is a vertical rectangle. I had no luck with the size or texture_size so will try changing its drag_rectangle properties. Can skip this for now as I think it wont matter when the tree size is set correctly to the map size.
• Create database (using SQLite3 for now?) and start playing with selecting a few from dropdown menus.

Create (or at least export) graphic assets in an editable opensource format, such as SVG (for vector) or GIMP (for raster).

all Adobe tools can generate SVG. Inkscape is a good opensource alternative.

I have no preference which, and usually basic pip works for me... but if you have multiple python projects with different environments needs at the same time you may prefer something else.

https://www.activestate.com/blog/which-python-dependency-manager-should-i-choose/

a few ones I like online:

• https://laforetcomestible.org/categorie/12-fruitiers-a-planter-le-plus-tot-possible/?pa_exposition=ombre

mobile UI has a few nice tricks, see

• https://www.search.io/blog/mobile-search-design-and-ux-best-practices
• https://www.search.io/blog/the-difference-between-filters-and-facets (esp: focus on emojis/icons and not text representation, so it's compact and international)

see appealing widgets such as https://www.pinterest.com/linimg/filter-multi-selection-ui/