Add Connect To Individual Overmind Processes Via tmux as a tmux TIL

README.md

@@ -10,7 +10,7 @@ working across different projects via [VisualMode](https://www.visualmode.dev/).
 
 For a steady stream of TILs, [sign up for my newsletter](https://visualmode.kit.com/newsletter).
 
-_1869 TILs and counting..._
+_1877 TILs and counting..._
 
 See some of the other learning resources I work on:
 
@@ -172,6 +172,7 @@ If you've learned something here, support my efforts writing daily TILs by
 - [Monitor Usage Limits From CLI](claude-code/monitor-usage-limits-from-cli.md)
 - [Open Current Prompt In Default Editor](claude-code/open-current-prompt-in-default-editor.md)
 - [Resume Specific Session](claude-code/resume-specific-session.md)
+- [Set Permission Mode When Starting Session](claude-code/set-permission-mode-when-starting-session.md)
 - [Stash The Current Prompt To Send Another First](claude-code/stash-the-current-prompt-to-send-another-first.md)
 
 ### Clojure
@@ -1105,6 +1106,7 @@ If you've learned something here, support my efforts writing daily TILs by
 - [Escape Curly Braces Within Formatted String](python/escape-curly-braces-within-formatted-string.md)
 - [Experiment With SQLite Queries In Memory](python/experiment-with-sqlite-queries-in-memory.md)
 - [Force Remaining Arguments To Be Named](python/force-remaining-arguments-to-be-named.md)
+- [Generate Sample PDFs With ReportLab](python/generate-sample-pdfs-with-reportlab.md)
 - [Get Absolute Seconds From `timedelta` Object](python/get-absolute-seconds-from-timedelta-object.md)
 - [Get Quotient And Remainder In One Operation](python/get-quotient-and-remainder-in-one-operation.md)
 - [Globally Install CLI Tool With UV](python/globally-install-cli-tool-with-uv.md)
@@ -1124,12 +1126,14 @@ If you've learned something here, support my efforts writing daily TILs by
 - [Parse Relative Time To datetime Object](python/parse-relative-time-to-datetime-object.md)
 - [Publish A Package To A Test Env As A Dry Run](python/publish-a-package-to-a-test-env-as-a-dry-run.md)
 - [Reclassify Certain Packages As Dev Dependencies](python/reclassify-certain-packages-as-dev-dependencies.md)
+- [Register SQLite Adapter To Serialize Datetimes](python/register-sqlite-adapter-to-serialize-datetimes.md)
 - [Resurface Exceptions Swallowed By Click Under Test](python/resurface-exceptions-swallowed-by-click-under-test.md)
 - [Select Implementation With Class Registry](python/select-implementation-with-class-registry.md)
 - [Set Up Pyright Type Checking In GitHub](python/set-up-pyright-type-checking-in-github.md)
 - [Skip Specific Pytest Test Cases](python/skip-specific-pytest-test-cases.md)
 - [Sort A List Of Dataclass Instances](python/sort-a-list-of-dataclass-instances.md)
 - [Sort Normalized Version Of Data](python/sort-normalized-version-of-data.md)
+- [Specify Parameter Name For Click Option](python/specify-parameter-name-for-click-option.md)
 - [Start Jupyter Notebook With Extra Packages](python/start-jupyter-notebook-with-extra-packages.md)
 - [Start The Debugger When A Test Errors](python/start-the-debugger-when-a-test-errors.md)
 - [Store And Access Immutable Data In A Tuple](python/store-and-access-immutable-data-in-a-tuple.md)
@@ -1648,9 +1652,11 @@ If you've learned something here, support my efforts writing daily TILs by
 
 ### SQLite
 
+- [Add Check Constraint To Existing Column](sqlite/add-check-constraint-to-existing-column.md)
 - [Display Results In Readable Column Format](sqlite/display-results-in-readable-column-format.md)
 - [Explore The Database Schema](sqlite/explore-the-database-schema.md)
 - [Manage Lightweight Schema Migrations With `user_version`](sqlite/manage-lightweight-schema-migrations-with-user-version.md)
+- [Output Query Result In Nicely Formatted Table](sqlite/output-query-result-in-nicely-formatted-table.md)
 
 ### Streaming
 
@@ -1680,6 +1686,7 @@ If you've learned something here, support my efforts writing daily TILs by
 - [Change Base Directory Of Existing Session](tmux/change-base-directory-of-existing-session.md)
 - [Change Base Directory Without Detaching](tmux/change-base-directory-without-detaching.md)
 - [Change The Default Prefix Key](tmux/change-the-default-prefix-key.md)
+- [Connect To Individual Overmind Processes Via tmux](tmux/connect-to-individual-overmind-processes-via-tmux.md)
 - [Create A Named tmux Session](tmux/create-a-named-tmux-session.md)
 - [Create A New Session In A New Server](tmux/create-a-new-session-in-a-new-server.md)
 - [Cycle Through Layouts](tmux/cycle-through-layouts.md)
@@ -1752,6 +1759,7 @@ If you've learned something here, support my efforts writing daily TILs by
 - [Check SSH Key Fingerprints Of Known Hosts](unix/check-ssh-key-fingerprints-of-known-hosts.md)
 - [Check The Current Working Directory](unix/check-the-current-working-directory.md)
 - [Check The Installed OpenSSL Version](unix/check-the-installed-openssl-version.md)
+- [Check What Is Inside A Zip File](unix/check-what-is-inside-a-zip-file.md)
 - [Clear The Screen](unix/clear-the-screen.md)
 - [Combine All My TILs Into A Single File](unix/combine-all-my-tils-into-a-single-file.md)
 - [Command Line Length Limitations](unix/command-line-length-limitations.md)

claude-code/set-permission-mode-when-starting-session.md

@@ -0,0 +1,29 @@
+# Set Permission Mode When Starting Session
+
+The way I typically use Claude Code day-to-day is with a couple long-running
+sessions for one to two clones of the project. I start a session with `claude`
+and then hit `shift+tab` until I've toggled it to _auto_ mode. I do tightly
+scoped features and `/clear` the context in between each.
+
+I get used to being in _auto_ mode, so whenever I start a new `claude` session I
+forget to first toggle from _manual_ to _auto_ mode.
+
+This is where the
+[`--permission-mode`](https://code.claude.com/docs/en/permission-modes) flag can
+help. I can start a session directly in _auto_ mode like so:
+
+```bash
+❯ claude --permission-mode auto
+```
+
+Or if I know I want to generate a plan first, I can start it in _plan_ mode.
+
+```bash
+❯ claude --permission-mode plan
+```
+
+There is also the `--dangerously-skip-permissions` flag which is equivalent to
+`--permission-mode bypassPermissions`. I tend to stay away from those unless I'm
+working from a sandboxed dev container.
+
+See `claude --help` for more details.

python/generate-sample-pdfs-with-reportlab.md

@@ -0,0 +1,44 @@
+# Generate Sample PDFs With ReportLab
+
+The `reportlab` package has a `pdfgen` module that provides a `Canvas` API for
+programmatically constructing a PDF.
+
+I recently needed the most basic use of this to quickly construct a sample PDF
+that I could use as I experimented with [different `qpdf`
+commands](/workflow/remove-pages-from-a-pdf.md). I wanted to generate a 10-page
+PDF where each page included a large font number in the center. This way I could
+clearly see what was the first, second, third, and so on pages. Then I could
+quickly verify that the `qpdf` manipulations I was doing were working as
+expected.
+
+I was able to do this without any setup or install. Using a `uv` one-liner (does
+a 10-line heredoc count as a one-liner?), I generated a PDF with some basic
+canvas rendering on each page.
+
+```python
+❯ uv run --with reportlab python - <<'EOF'
+from reportlab.pdfgen import canvas
+N, W, H = 10, 612, 792
+c = canvas.Canvas("test.pdf", pagesize=(W, H))
+for i in range(1, N + 1):
+    c.setFont("Helvetica-Bold", 220)
+    c.drawCentredString(W / 2, H / 2 - 80, str(i))
+    c.setFont("Helvetica", 28)
+    c.drawCentredString(W / 2, 100, f"page {i} of {N}")
+    c.showPage()
+c.save()
+EOF
+```
+
+The `--with` makes the `reportlab` package available to the command being run.
+The command being run is `python` which evaluates that block of heredoc text.
+
+Within the heredoc, I import `canvas` from the `reportlab` package that has been
+made available. I then create an in-memory PDF of a standard size. Then for-loop
+10 times to build each page of the PDF with a big number in the middle and a
+smaller `page x of N` lower down on the page. Finally, I save the PDF which
+writes it to the directory I executed this whole command from.
+
+I had Claude help with the `canvas` API, however there are some decent
+[`reportlab.pdfgen` docs
+here](https://docs.reportlab.com/reportlab/userguide/ch2_graphics/).

python/register-sqlite-adapter-to-serialize-datetimes.md

@@ -0,0 +1,93 @@
+# Register SQLite Adapter To Serialize Datetimes
+
+SQLite doesn't have a datetime or timestamp data type, so that information has
+to be stored as `text` or a unix epoch `int`. That means when using Python's
+`sqlite3` module to perform writes I need to tell it what shape to write a
+datetime value.
+
+I could manually convert `datetime` values everywhere they are involved in
+_write_ operations.
+
+```python
+# Prepare `sessions` insert payload
+session_data = {
+    "active": 1 if active else 0,
+    "project_id": project_id,
+    "start_time": datetime.isoformat(session.start_time),
+    "end_time": None,
+}
+
+if session.end_time:
+    session_data["end_time"] = datetime.isoformat(session.end_time)
+
+# Insert the new active session
+cursor = self.conn.execute(
+    """
+    insert into sessions (active, project_id, start_time, end_time)
+    values (:active, :project_id, :start_time, :end_time)
+    returning id;
+""",
+    session_data,
+)
+```
+
+I've used `datetime.isoformat` above which formats `datetime` objects like so:
+
+```python
+>>> datetime.now().isoformat()
+'2026-08-28T11:52:04.709907'
+```
+
+I'd like to make two improvements.
+
+1. I want these `datetime` values to be formatted instead like `2026-08-28T18:15:27.213Z`.
+2. I want `datetime` values to be serialized automatically in the specific shape
+   without having to manually convert them everywhere.
+
+I can achieve both of these things by [registering an adapter with `sqlite3`](https://docs.python.org/3/library/sqlite3.html#how-to-register-adapter-callables)
+that handles the serialize of `datetime` objects.
+
+First, I define a function that can perform the `datetime` to `str` conversion.
+I decided to put this in `db.py` where I have some other database-specific
+functions.
+
+```python
+from datetime import datetime, timezone
+
+def to_db(dt: datetime) -> str:
+    if dt.tzinfo is None or dt.utcoffset() is None:
+        raise ValueError(f"Unable to store naive datetime: {dt!r}")
+    dt = dt.astimezone(timezone.utc)
+    return f"{dt:%Y-%m-%dT%H:%M:%S}.{dt.microsecond // 1000:03d}Z"
+```
+
+Then I register the adapter before creating the connection that gets used for
+database interactions.
+
+```python
+import sqlite3
+from datetime import datetime
+from pathlib import Path
+from sqlite3 import Connection
+
+def initialize_conn(db_file: Path) -> Connection:
+    # register adapters
+    sqlite3.register_adapter(datetime, to_db)
+
+    conn: Connection = sqlite3.connect(db_file)
+    conn.row_factory = sqlite3.Row
+
+    return conn
+```
+
+Then I can run write operations with `datetime` objects knowing they will be
+correctly serialized.
+
+```python
+with self.conn:
+    query = "update sessions set active = :active, end_time = :end_time where active = 1;"
+    self.conn.execute(
+        query,
+        {"active": 0, "end_time": session.end_time},
+    )
+```

python/specify-parameter-name-for-click-option.md

@@ -0,0 +1,55 @@
+# Specify Parameter Name For Click Option
+
+[Click](https://click.palletsprojects.com/en/stable/)'s option decorator
+provides a versatile way to define flags for a command. It has good defaults
+that minimize the aspects of a flag that I need to be explicit about.
+
+For example, a boolean `--init` flag for the `config` command could be specified
+like so:
+
+```python
+@cli.command()
+@click.option(
+    "--init",
+    help="Initialize a config file with minimal defaults",
+    is_flag=True,
+)
+@pass_cli
+def config(cli_ctx: CliContext, init: bool):
+    # ...
+```
+
+Notice, in particular, that the flag string (`--init`) that I pass as the first
+argument to `@click.option` has to correspond to the name of the parameter
+`init`. Click passes all the defined options as keyword arguments when invoking
+`config`. If the `init` parameter was changed to `initial`, there would be a
+runtime error like this: `TypeError: config() got an unexpected keyword argument 'init'`.
+
+Like I said though, Click is flexible when I need it to be. I can leave the flag
+name as it is, but specify a different name to be used for the function
+parameter. I found this useful when I realized that as I added support for a
+`--json` flag I was inadvertently superseding the `json` import.
+
+The second positional argument to `@click.option` can be included to rename that
+parameter:
+
+```python
+import json
+
+
+@cli.command()
+@click.option(
+    "--json",
+    "use_json",
+    help="Output all info details in JSON format",
+    is_flag=True,
+)
+@pass_cli
+def info(cli_ctx: CliContext, use_json: bool):
+    # ...
+
+    if use_json:
+        click.echo(json.dumps(info_details, indent=2))
+```
+
+Both of these code blocks are excerpts from my [`py-vmt` time tracker project](https://github.com/jbranchaud/py-vmt).

sqlite/add-check-constraint-to-existing-column.md

@@ -0,0 +1,74 @@
+# Add Check Constraint To Existing Column
+
+To not bury the lede, SQLite does not support adding a check constraint to an
+existing column. Instead, if I want to add one after the fact, I can reconstruct
+the table with the check constraint specified up front and then migrate the
+data.
+
+Here is a `sessions` table that includes `start_time` and `end_time` columns
+that use the `text` data type to represent points in time.
+
+```sql
+sqlite> .schema sessions
+CREATE TABLE sessions (
+        id integer primary key,
+        active integer not null check (active in (0, 1)),
+        project_id integer not null references projects(id) on delete cascade,
+        start_time text not null,
+        end_time text,
+        created_at text not null default (datetime('now')),
+        updated_at text not null default (datetime('now'))
+    );
+CREATE UNIQUE INDEX idx_sessions_single_active
+        on sessions(active)
+        where active = 1;
+```
+
+I want `start_time` and `end_time` to both enforce the shape of the timestamp
+strings with `check` constraints. To do that, I need to start a transaction,
+create a new version of the table with the check constraints, migrate the data,
+rename `sessions` to `sessions_old`, rename `sessions_new` to `sessions` (that's
+the in-place swap), and then commit the transaction. `sessions_old` can be
+dropped later once I feel good about the migration.
+
+```sql
+begin transaction;
+
+create table sessions_new (
+  id integer primary key,
+  active integer not null check (active in (0, 1)),
+  project_id integer not null references projects(id) on delete cascade,
+  start_time text not null,
+  end_time text,
+  created_at text not null default (datetime('now')),
+  updated_at text not null default (datetime('now')),
+  check(start_time is strftime('%Y-%m-%dT%H:%M:%fZ', start_time)),
+  check(end_time is strftime('%Y-%m-%dT%H:%M:%fZ', end_time))
+);
+
+insert into sessions_new (
+  id, active, project_id, start_time, end_time, created_at, updated_at
+)
+select id, active, project_id, start_time, end_time, created_at, updated_at
+from sessions;
+
+alter table sessions rename to sessions_old;
+
+alter table sessions_new rename to sessions;
+
+drop index if exists idx_sessions_single_active; -- on sessions_old
+
+create unique index idx_sessions_single_active
+  on sessions(active)
+  where active = 1;
+
+commit;
+```
+
+Notice that after the table renames I also drop index (now pointing to
+`sessions_old`) and recreate it for the _new_ `sessions` table.
+
+This approach worked well for my situation, but may not be a one-size-fits-all
+solution. Depending on how the database is deployed, the size of the tables, and
+usage, this approach may not scale. Always do a dry-run of database migrations
+like this.

sqlite/output-query-result-in-nicely-formatted-table.md

@@ -0,0 +1,50 @@
+# Output Query Result In Nicely Formatted Table
+
+When I start a fresh SQLite connection and run a query, all the results are
+squished together in a way that is poorly formatted, hard to read, and missing
+column headers.
+
+```sql
+sqlite> select sessions.id, start_time, end_time, projects.name from sessions join projects on projects.id = sessions.project_id limit 3;
+1|2026-07-26T21:15:50.062936+00:00|2026-07-26T21:53:13.990Z|taco
+2|2026-07-26T21:53:40.019946+00:00|2026-07-26T22:14:09.168Z|TIL
+3|2026-08-02T17:00:16.119169+00:00|2026-08-02T17:30:16.119Z|py-vmt
+```
+
+I can drastically improve the look of this by turning _headers_ on and switching
+to _box_ mode.
+
+```sql
+sqlite> .headers on
+sqlite> .mode box
+sqlite> select sessions.id, start_time, end_time, projects.name from sessions join projects on projects.id = sessions.project_id limit 3;
+┌────┬──────────────────────────────────┬──────────────────────────┬────────┐
+│ id │            start_time            │         end_time         │  name  │
+├────┼──────────────────────────────────┼──────────────────────────┼────────┤
+│ 1  │ 2026-07-26T21:15:50.062936+00:00 │ 2026-07-26T21:53:13.990Z │ taco   │
+│ 2  │ 2026-07-26T21:53:40.019946+00:00 │ 2026-07-26T22:14:09.168Z │ TIL    │
+│ 3  │ 2026-08-02T17:00:16.119169+00:00 │ 2026-08-02T17:30:16.119Z │ py-vmt │
+└────┴──────────────────────────────────┴──────────────────────────┴────────┘
+```
+
+I personally find that much easier on the eyes. It is also a nicer format to
+copy and paste into a post like this or a formatted code block that I'm sharing
+with a colleague.
+
+I can also do this directly from the CLI with a one-liner using `-header` and
+`-box` like so:
+
+```bash
+❯ sqlite3 /Users/lastword/.local/share/vmt/sessions.db -header -box "select sessions.id, start_time, end_time, projects.name from sessions join projects on projects.id = sessions.project_id limit 3"
+┌────┬──────────────────────────────────┬──────────────────────────┬────────┐
+│ id │            start_time            │         end_time         │  name  │
+├────┼──────────────────────────────────┼──────────────────────────┼────────┤
+│ 1  │ 2026-07-26T21:15:50.062936+00:00 │ 2026-07-26T21:53:13.990Z │ taco   │
+│ 2  │ 2026-07-26T21:53:40.019946+00:00 │ 2026-07-26T22:14:09.168Z │ TIL    │
+│ 3  │ 2026-08-02T17:00:16.119169+00:00 │ 2026-08-02T17:30:16.119Z │ py-vmt │
+└────┴──────────────────────────────────┴──────────────────────────┴────────┘
+```
+
+Run the `.help` dot-command from a SQLite prompt for a full listing of these
+commands. See also `sqlite3 --help` from the CLI for usage details about all
+flags.

tmux/connect-to-individual-overmind-processes-via-tmux.md

@@ -0,0 +1,39 @@
+# Connect To Individual Overmind Processes Via tmux
+
+A common `Procfile.dev` in a Rails project might look something like this.
+
+```
+web: bin/rails server -p $PORT
+vite: bin/vite dev
+worker: bundle exec good_job start
+```
+
+Instead of starting up each process that needs to be running for development to
+work, I can instead run a tool that reads the procfile and sets it all up for me
+-- like `overmind`.
+
+```bash
+❯ overmind start -f Procfile.dev
+```
+
+What's cool about `overmind` is that it starts its own `tmux` session and then
+runs each of these processes in its own window.
+
+I can connect to any one of them by name with `overmind connect <name>`. Or I
+can connect to the session defaulting to the first window with just `overmind
+connect`.
+
+If I need to see what is going on with my background jobs, I'll run:
+
+```bash
+❯ overmind connect worker
+```
+
+This behaves like any other tmux session, so I can use my prefix key (`ctrl-z`
+in my case) to access tmux-specific keybindings. Most notably, once I'm done
+looking, I'll want to hit `ctrl-z d` to detach from the session.
+
+I'm already using tmux as my daily driver which means its easy for me to end up
+in a nested tmux session if I connect while already in my development session.
+To help with that, I set up [a forwarding
+prefix](set-up-forwarding-prefix-for-nested-session.md).

unix/check-what-is-inside-a-zip-file.md

@@ -0,0 +1,31 @@
+# Check What Is Inside A Zip File
+
+Before unzipping a file, it can be useful to check the contents. You can learn a
+few things from doing this. First off, are all the files nested under some extra
+top-level directory? How many files are in there? How big are certain files? Etc.
+
+The `-l` flag to `unzip` can help answer these questions.
+
+```bash
+❯ unzip -l zipped-files.zip
+Archive:  zipped-files.zip
+  Length      Date    Time    Name
+---------  ---------- -----   ----
+        0  08-22-2026 08:54   files-to-zip/
+       14  08-22-2026 08:54   files-to-zip/data.json
+       38  08-22-2026 08:53   files-to-zip/notes.md
+       14  08-22-2026 08:53   files-to-zip/hello.txt
+---------                     -------
+       66                     4 files
+```
+
+Notice that the first line is for `files-to-zip/` which tells us that everything
+is nested under the top-level directory that was probably the target of the
+`zip` command. Everything is organized into a table with labeled columns which
+allows us to answer our questions and get a good idea of what is in there.
+
+In my experience, zip files tend to have dozens to hundreds of files, so it is
+even better to pipe the above to a pager like `less`. That way you can search
+and navigate through everything that's in there.
+
+See `man unzip` for more details.