snap_compare failures using DirectoryTrees and .mount in spite of identical appearance

[mitaa]

The MRE below always generates the same appearance in the terminal and snapshot_report.html,
but creates non-identical svg files leading to snap_compare failures (well sometimes, sometimes not).

Seemingly since textual 0.80

I've created a small git repo with the MRE and svg test samples here:
https://github.com/mitaa/mre_textual_svg_dtree

After taking 20 samples with each version i get the following

(filename = hash)

├───0.79.0
│       103ed17be2726a686e7b85e77fa2ab03af85c344aeb012985cda803a8147403d.svg
│
├───0.80.0
│       66ba2bb8610dfac2b1e8ad668ef79dd7b5e2636713fb586f4350e744ba0c2648.svg
│       8cbe0a12044fdeb2c5d763cc079c816d65ab9fda51764a1c498e1f8b9e74414b.svg
│
└───0.81.0
        2769b26e6544461cc1d3d5c6344a64b9353f7e6ef6fcd80417dfb929dfa79089.svg
        924cf504886288b5587dc86f2e517aaa5e8d89e7abd320f72a6afef013ecb8d9.svg

The difference within the svg files seem to be the UIDs (.terminal-) here
2769b26e6544461cc1d3d5c6344a64b9353f7e6ef6fcd80417dfb929dfa79089.svg

    .terminal-2464235620-matrix {
        font-family: Fira Code, monospace;
        font-size: 20px;
        line-height: 24.4px;
        font-variant-east-asian: full-width;
    }

924cf504886288b5587dc86f2e517aaa5e8d89e7abd320f72a6afef013ecb8d9.svg

    .terminal-2683650149-matrix {
        font-family: Fira Code, monospace;
        font-size: 20px;
        line-height: 24.4px;
        font-variant-east-asian: full-width;
    }

MRE code from the repo:
mre.py

from textual.app import App
from textual.widgets import DirectoryTree

ROOT = r"foo"


class MyApp(App):
    # This works
    # def compose(self):
    #     yield DirectoryTree(ROOT)
    #     yield DirectoryTree(ROOT)

    # This doesn't
    def on_mount(self) -> None:
        self.mount(DirectoryTree(ROOT))
        self.mount(DirectoryTree(ROOT))

    def on_tree_node_highlighted(self, node):
        self.query(DirectoryTree)[-1].reload()


def main():
    app = MyApp()
    app.app.run()


if __name__ == "__main__":
    main()

sample.py

import os
from pathlib import Path
from hashlib import sha256
import shutil

import textual
from textual._doc import take_svg_screenshot

from mre import MyApp


def main():
    root = Path("test_results") / textual.__version__
    os.makedirs(root, exist_ok=True)

    for _ in range(20):
        app = MyApp()
        actual = take_svg_screenshot(app).encode("utf-8")

        fpath = root / f"{sha256(actual).hexdigest()}.svg"
        fpath.write_bytes(actual)

if __name__ == "__main__":
    main()

Textual Diagnostics

Versions

Name	Value
Textual	0.81.0
Rich	13.8.1

Python

Name	Value
Version	3.12.6
Implementation	CPython
Compiler	MSC v.1940 64 bit (AMD64)
Executable	C:\Program Files\Python312\python.exe

Operating System

Name	Value
System	Windows
Release	10
Version	10.0.19045

Terminal

Name	Value
Terminal Application	Unknown
TERM	xterm-256color
COLORTERM	truecolor
FORCE_COLOR	Not set
NO_COLOR	Not set

Rich Console options

Name	Value
size	width=170, height=46
legacy_windows	False
min_width	1
max_width	170
is_terminal	True
encoding	utf-8
max_height	46
justify	None
overflow	None
no_wrap	False
highlight	None
markup	None
height	None

[github-actions]

We found the following entries in the FAQ which you may find helpful:

• How do I center a widget in a screen?
• Why doesn't Textual look good on macOS?
• Why do some key combinations never make it to my app?

Feel free to close this issue if you found an answer in the FAQ. Otherwise, please give us a little time to review.

This is an automated reply, generated by FAQtory

[mitaa]

Correction:

    # This works
    # def compose(self):
    #     yield DirectoryTree(ROOT)
    #     yield DirectoryTree(ROOT)

Not entirely.
Using this with 0.81 seems to generate identical svg files, but they are still different from other versions.

Using the code quoted above instead of the on_mount function i get the following

├───0.78.0
│       103ed17be2726a686e7b85e77fa2ab03af85c344aeb012985cda803a8147403d.svg
│
├───0.79.0
│       103ed17be2726a686e7b85e77fa2ab03af85c344aeb012985cda803a8147403d.svg
│
├───0.80.0
│       66ba2bb8610dfac2b1e8ad668ef79dd7b5e2636713fb586f4350e744ba0c2648.svg
│       8cbe0a12044fdeb2c5d763cc079c816d65ab9fda51764a1c498e1f8b9e74414b.svg
│
└───0.81.0
        2769b26e6544461cc1d3d5c6344a64b9353f7e6ef6fcd80417dfb929dfa79089.svg

[willmcgugan]

Snapshots may fail when you update Textual. This can happen when we change something. More often than not there is no visual change to the output, and you can just accept the changes.

If you always get the same output for a particular version of Textual, then that indicates there is no problem -- assuming the output is what you expected.

If you are getting different output which the same version of Textual, then that most often indicates that the output is being generated differently run-to-run.

From your code I can see you aren't awaiting the mounts or the reload. If you don't await them, they will run at the same time, likely introducing a race condition.

Your sample.py is not using our test framework. Which makes this less predictable. I would suggest testing via a pytest test. If you don't see issues after awaiting the tree operations, then that may be the fix.

[mitaa]

You're totally right, awaiting them makes my issue disappear.

Thanks and sorry about that.

[github-actions]

Don't forget to star the repository!

Follow @textualizeio for Textual updates.

[willmcgugan]

Not a problem!