Satis, I am confused. Please help me understand better.

After reading your post several times, here is my understanding of what you are describing. A child row is always created when hitting Return after typing some text. You expect a sibling row to be created, but a child row is created instead.

In Zavala (and Bike AFAIK) the only time a child row is automatically created is when you are at the end of a row and that row already has child rows. In that case it will insert a new child row above the other child rows.

I'm having trouble reconciling what I think you are saying verses how Zavala actually works.

Maurice Parker wrote:

Satis, I am confused. Please help me understand better.

After reading your post several times, here is my understanding of what
you are describing. A child row is always created when hitting Return
after typing some text. You expect a sibling row to be created, but a
child row is created instead.

In Zavala (and Bike AFAIK) the only time a child row is automatically
created is when you are at the end of a row and that row already has
child rows. In that case it will insert a new child row above the other
child rows.

I'm having trouble reconciling what I think you are saying verses how
Zavala actually works.

In Zavala, when I’m editing a row that already has child rows, pressing Return doesn’t create a new sibling row (which is what I’ve come to expect). Instead, it creates a new child row. That means I then have to press Control–Command–Left Arrow to outdent it, something I was having to do fairly often. As far as I can recall, there isn’t a direct keyboard command in Zavala to simply create a new sibling row in this situation.

I often return to existing rows to edit them, and when those rows have children, I almost never want Return to create a new child item. Much more often, I want to continue adding items at the same outline level I was already working in. This expectation is common in many non-outline apps. For example, in Reminders or Todoist, when I edit a task and press Return, I expect a new task to be created -- not to be dropped into the task’s notes section (entering notes requires a different command). Likewise, in Excel and Google Sheets, pressing Return (or Enter) finishes editing a cell and moves to the next one, rather than inserting a line break inside the cell (as Apple Numbers does, sometimes to the surprise or frustration of users accustomed to the more common behavior).

I should have been clearer that Zavala does not continue to indent with additional Returns; I was describing in general the consequence of always adding new children with a Return, which I've seen in other apps.

Thanks for the clarification, Satis. I'll definitely do my best to get this in the 4.0 release. I hate making promises about upcoming features, but I really can't see how I can't make this happen for 4.0.

The issue that I created to track this is here: https://github.com/vincode-io/Zavala/issues/298

Zavala 4.0 is now in Beta testing. Zavala 4.0 has been updated to support Liquid Glass and requires the latest OS's from Apple to work. It also has the new Setting so that the Editor can create sibling rows instead of child rows when return is used on a row with child rows. Internally, there were some big changes to make syncing more robust. Unfortunately that required updating the internal database and the iCloud database schemas. That means that once you upgrade to Zavala 4.0, you can't go back to a prior version and if you are using iCloud syncing, all your devices have to be upgraded to Zavala 4.0 to sync using the new database structure.

Any help testing Zavala 4.0 is greatly appreciated. Here is more on how you can help test.

https://github.com/vincode-io/Zavala/discussions/300

It looks very nice, but I'm very wary of software where I can't see where my files are, or *touch* them.

The OPML import is neat, but I'd love to see a first-class citizen of OPML, where it doesn't get imported and exported, but can just be opened and saved as a normal file.

But the 'magic storage' of the files is a killer for me: MindNode has gone this way in Next, and it means I can't use it reliably as the file doesn't exist on my machine and can't be copied, rendered by other application - it can only be queried through scriping in other tools, which turns it into a dead-end silo.

/justmythoughtsonopening

Totally agree, although I have a little tolerance for proprietary file formats in some cases.

If I can understand enough of the format to extract my work, that's a little bit of a safety net.

The worst, for me, is not being able to share specific documents with a second machine. I try to avoid the cloud, so individual documents I can carry around on an usb drive are my preference.

I stopped using TheBrain long ago. At one point I was tempted to reinstall but it puts everything you do in one database. No more individual Brain files.

Dave M wrote:

It looks very nice, but I'm very wary of software where I can't see
where my files are, or *touch* them.

The OPML import is neat, but I'd love to see a first-class citizen of
OPML, where it doesn't get imported and exported, but can just be opened
and saved as a normal file.

But the 'magic storage' of the files is a killer for me: MindNode has
gone this way in Next, and it means I can't use it reliably as the file
doesn't exist on my machine and can't be copied, rendered by other
application - it can only be queried through scriping in other tools,
which turns it into a dead-end silo.

/justmythoughtsonopening

Zavala 4.0 is now up in the App Store.
https://apps.apple.com/us/app/zavala/id1546457750

This release is a moderate one in features even though it took a fair amount of work. I did get Satis' issue with how rows with children get created. For more on that and the other features in this release, I wrote a blog post called Building Zavala 4.0.
https://vincode.io/2026/02/16/building-zavala.html

Looks great, Maurice.

In the "Need a memory jog" thread there was some discussion of AI and using it in a mind mapping application. I've always considered mind mappers and outliners to be cousins. Anyway it got me thinking about how AI relates to Zavala. I wrote a blog post about it.

https://vincode.io/2026/02/28/zavala-ai.html

If you don't want to read the whole post, I basically say that I'm not including an AI features in Zavala anytime soon. I also show you how to work around that using Shortcuts. If you are a Bike or other outliner user, you might want to check it out too. The demo Shortcut I created should work with Bike or any other outline that can import OPML files using Shortcuts.

AI paralyzes me. I know in my gut we're going to see a stock market hit at least as big as the housing bubble. That took out half the market's value.

I'm paralyzed because I can't bring myself to close my retirement investments. They are tiny and I can't afford to lose a penny. Worse, if I close them I have to time it to avoid losses bigger than the IRS penalty for sidestepping disaster.

Sounds quite alarmist, but look at Oracle. It hit a peak in September. By January it lost half of that from an AI contract that fell through.

There's a data center being built adjacent to Hubbard, Texas. The town is 1200 acres. The datacenter will cover 2000.

The new DC will have 600 MW of generating capacity with on site natural gas turbines. By my calculations, that will burn about 2% of the total gas production in Texas in 2025. Add the much larger Fermi data center in Amarillo and Chevron's huge project in Pecos County and that comes to nearly half the state's gas production - from just those three data centers.

Hubbard's population is about 1200. 600 MW will run about 288,000 homes.

If you support AI, you have to have hope AI will be here for the long run. It probably will, but not in its present form.

OpenAI will lose billions this year while committing to the better part of a trillion in construction. I think new chips will emerge using a fraction of the power of Nvidia's current offering. Great, we can turn back the throttle on power generation but we can't throttle back the debt obligations funding future overcapacity. New competitors will sideline the current generation, whose bankruptcies will avalanche into the economy at large.

Good luck to us all.

I understand the gut-based anxiety but the math is off.

Texas produced roughly 34.4 Bcf/day (~12,556 Bcf/year) of natural gas in 2025. A 600 MW gas plant running flat out all year would consume about 46 Bcf/year, roughly 0.37% of Texas production, not 2%. Even if you assume the three major projects all run at their full theoretical gas capacity simultaneously:

Nexus (Hubbard) – 600 MW: ~46 Bcf (~0.4%)

Fermi (Amarillo) – 6 GW gas approved: ~464 Bcf (~3.7%)

GW Ranch (Pecos County) – 7.65 GW: ~591 Bcf (~4.7%)

Combined, that’s under 9% of Texas annual not nearly half. And that assumes continuous full-capacity operation, which is rather unlikely. So the gas-consumption premise driving the broader collapse argument doesn’t hold up.

On the investing side crashes absolutely happen (and people should expect them, along with recesssions) but note that the 2008 crash recovered from its *peak* in 5–6 years. The bigger risk for most people isn’t a crash; it’s trying to time exits and re-entries based on fear. People are terrible at market timing. And people shouldn't expect the stock market to be perfectly safe (though medium- and long-term it has proven to be... so far).

Single-stock volatility doesn’t imply systemic collapse. Oracle stock surged then dropped sharply on a single contract loss (actually it was a combination of things including a weak earnings guidance and soaring AI costs as well), but today it's just past where it was a year previous. That's volatility, not structural collapse.

Tech cycles often overbuild (fiber in 2000 is a good example whch I've used on this site). Some firms fail, but infrastructure remains and becomes cheaper and more productive. Bankruptcies don’t automatically equal economic collapse. The dot-com telecom collapse of 2000–2002 produced one of history's greatest accidental infrastructure gifts: a planet-spanning fiber network built at enormous cost, left largely unused after bankruptcy, and ultimately repurposed to power the modern internet, cloud computing, and AI.

Arg. You're right. In my defense, I didn't use a calculator and slipped a few decimal places.

Many thanks for the correction.

Unless I goofed again, Hubbard will still burn 158 billion cubic meters of gas per year, assuming they are generating the full 600 MW. I don't know if 600 MW is the expected load or the designed generating capacity.

Seems a poor use of natural resources either way.

Zavala 4.1 is released. It took quite a while to get it through App Store review. I suspect that AI coded projects are overwhelming Apple''s app review process. Anyway this release adds Markdown import and some new formatting options. The Markdown import works better than I thought it could. Shoe horning a mostly unstructured format into an outline is challenging and is an imperfect process, but it works. More on that here:

https://vincode.io/2026/03/17/zavala.html

I've already got the 4.2 release done and I am testing it now. It adds encryption and a couple new Shortcut actions. It should be a solid release.

Maurice Parker wrote:

I suspect that AI coded projects are overwhelming Apple''s
app review process.

557,000 new iOS apps were submitted in 2025, which is +24% year-over-year growth, after multiple years of flat or even declining new app output. Up to 2,000 new apps per day. AI is likely the main driver for it.

Still, that's below Apple's peak, which I think was around 1 million apps in 2016. But this recent growth is probably a shock to their evaluation system, a system which is responsible for checking a lot more for security (and payments) than a decade ago.

The Markdown import works better than I thought it could.

Very cool.

Zavala 4.2 is out. This release adds Outline Locking. This functions basically the same as how Apple's Notes does locking. It encrypts Outline contents on the filesystem and while syncing using iCloud.

Looking forward to Zavala 4.3 and I'm not sure what to include. I'd be interested in hearing suggestions as long as it isn't direct access to the underlying datastore. :-) I realize that is a popular feature with CRIMPers, but that probably won't ever happen.