Streamlit · 08 — FAQ: Common Questions and Gotchas¶

Answers to the questions beginners actually ask. If your app misbehaves, read this list before searching the internet.

1. Why does my app run the whole script every time I click something?¶

That's the design. Streamlit re-runs the script top to bottom on every interaction (a "rerun"). It's what makes apps so easy to write — your code is always in sync with the widget values. The cost is that expensive work repeats, which is why caching (@st.cache_data) exists. Once you internalize "rerun", most Streamlit surprises disappear.

2. My button doesn't "remember" anything. Why?¶

Buttons are True only for the single rerun caused by the click. If you store the result in a normal variable, it's gone on the next rerun. Use st.session_state for anything that must survive across reruns:

if "count" not in st.session_state:
    st.session_state.count = 0

3. What exactly does `st.session_state` do?¶

It's a per-user, per-session dictionary that persists across reruns. Use it for counters, chat history, choices that should stick. Do not use it for expensive computation — that's what caching is for.

4. `@st.cache_data` — when does it actually run my function?¶

The first time you call it with a given set of arguments. If the arguments are the same on the next rerun, the stored result is returned instead. Change an argument → new computation, new cache entry. That's why caching "doesn't work" when a widget value is one of the arguments — every new value is a new cache key. Cache on data, not on widgets.

5. My chart is empty. Why?¶

Most common causes: (a) the dataframe is empty after filtering — check len(df) and show st.warning(...); (b) you forgot st.pyplot(fig); (c) seaborn drew into the wrong axes — always fig, ax = plt.subplots() and pass ax=ax to seaborn.

6. My app shows an error/red text. What do I do?¶

Read the red traceback in the app — it's the real Python error, usually a column name, a division by zero, or a missing value. Then check the "Common mistakes" table in the relevant document. To see the full traceback, scroll down in the app or run the script directly (python app.py won't run Streamlit, but the pure-Python parts will run and often reveal the error).

7. How do I stop an app? How do I change the port?¶

Ctrl+C in the terminal stops it. If port 8501 is busy: streamlit run app.py --server.port 8502.

Yes, two levels: - Same network: run streamlit run app.py --server.address 0.0.0.0 and share your machine's IP — classmates on the same Wi-Fi can open it. - Public: deploy to a free hosting service (Streamlit Community Cloud, Render, Hugging Face Spaces). Basic deployment is covered in document 7; the rule for this course: never deploy an app containing private data — public hosting means public data.

9. Why is `st.file_uploader` giving me a weird object instead of a DataFrame?¶

It returns a file-like object, not a DataFrame. Convert with pandas: df = pd.read_csv(uploaded). Check uploaded is not None first — on the first run nothing is uploaded yet.

10. My `multiselect` crashes when I select nothing.¶

st.multiselect returns an empty list when nothing is selected, and df[[]] errors. Guard it:

cols = st.multiselect("Columns", df.columns, default=df.columns[:2])
if cols:                    # empty list is falsy
    st.dataframe(df[cols])
else:
    st.info("Pick at least one column.")

11. Widgets with the same label conflict. Why?¶

Streamlit identifies widgets by label. Two widgets with the same label on the same page collide. Fix: give them unique labels, or add key=: st.selectbox("Day", days, key="eda_day").

12. My app is slow. Where do I start?¶

Three usual suspects, in order: (1) the data is reloaded on every rerun → cache the load; (2) the model retrains on every rerun → cache the training; (3) you're plotting the full dataframe → downsample or aggregate before plotting.

13. What's the difference between `st.dataframe` and `st.table` again?¶

st.dataframe: interactive — sortable, scrollable, searchable; use for real data exploration. st.table: static; use for small fixed results (top-5, a summary). See document 4.

14. Why doesn't `print()` show anything?¶

print goes to the terminal where streamlit run is executing, not to the app page. To see values in the app, use st.write(...).

15. Do I need to know HTML/CSS/JavaScript for this course's apps?¶

No. The whole module is pure Python. (If you later deploy to a custom domain you'll meet config files, but nothing in this course requires web code.)

16. How do I add an AI chatbot to my Streamlit app?¶

See Module C topic 11 (../module-c/topic-11-ai-in-streamlit.md): st.chat_input + st.session_state for history + a local Ollama call. The short version: the chat history must live in session state, and the LLM call is just a function the app runs on rerun.

17. My deployed app shows the wrong data / leaks data.¶

Never deploy an app that reads private files or hardcodes credentials. Load data from a public source or let users upload it (the uploader pattern from document 4 keeps data off the server entirely).

18. Where is the full reference?¶

streamlit-cheatsheet.md (this module) and the official docs: https://docs.streamlit.io — search any st. function name.