Make it easy to print to the app!

[codeananda]

Yesterday was my first day using wave and I spent 5 hours pulling my hair out and feeling very confused.

One thing I found particularly frustrating was that you cannot easily print info directly to the app. If you run print() from your .py file it prints to the terminal which gets clogged with loads of other output so it's hard to tell what is happening.

For example, when loading a dataframe, it would be nice if I could run df.head() or even print('dataframe loaded successfully'). All DSs/MLEs do this countless times per day in jupyter notebooks... but it is very unintuitive in wave. For example, in streamlit anything written on its own is automatically printed to the app. Plus, they have the write() function/method. This is one of the first things they cover in their docs and makes debugging your apps an intuitive experience.

What is the recommended way to do this with wave? Adding this to your docs would be super helpful.

[mturoci]

Hi @codeananda, you can set H2O_WAVE_NO_LOG. This way you can see your print statements easier. See configuration docs for more useful features.

What is the recommended way to do this with wave? Adding this to your docs would be super helpful.

You can use any logging mechanism as you wish. See our docs for an example.

[azim-b]

Hi @codeananda, you can also write logs to a file, and view the tail of a file with -f option so that you only see your logs and the tail is automatically updated.

Ex: tail -f my_wave_app.log

Also, like @mturoci mentioned, setting H2O_WAVE_NO_LOG to True was quite helpful for me.

[mtanco]

@codeananda When you say "printed to the app" do you mean you would like the content of df.head() to appear not in the terminal for the app developer to debug but rather in the App itself for all end users to see?

Based on the answer to this I have some suggestions, but I want to make sure we are understanding your request

[codeananda]

@mtanco I meant printing to the app itself. One reason for this is that I'm using plotly and this makes using the terminal for debugging impossible (see #1374 I just opened).

I've also been developing locally so wasn't worrying about other people seeing what I push to the app.

I'm inspired by Streamlit with their magic and st.write() function.

@mturoci thanks for the H2O_WAVE_NO_LOG point. Alas, it didn't solve my problem due to #1374.

@azim-b Being honest, I've never used logging before (probably a DS/ML sin, I know!) so it feels much less natural/overkill to do this than printing (especially when checking if df has loaded). I imagine many other devs will be in my shoes... though perhaps we just need to bite the bullet and use logging!

[mtanco]

It would need to be custom, since in general Wave is focused on building web apps, but here are some functions for "logging" to your screen:

from h2o_wave import main, app, Q, ui
import pandas as pd


@app('/')
async def serve(q: Q):
    create_log_content(q)

    wave_log(q, "testin")
    wave_log(q, "testin moar")
    df = pd.DataFrame([1, 1, 2], columns=["test"])
    wave_log(q, df)
    print(df)
    await q.page.save()


def create_log_content(q: Q):
    # Each browser tab sees only their own logs
    q.client.log_content = "# My Logs\n"
    q.page["logs"] = ui.markdown_card(
        box="10 1 -1 -1",
        title="",
        content=q.client.log_content
    )


def wave_log(q: Q, content):
    content = str(content)
    content = content.replace("\n", "\n\n")
    content += "\n\n*---------*\n\n"
    q.client.log_content += content
    q.page["logs"].content = q.client.log_content