preview 5; 2.0 release plans ?

[Simon Michael]

Hi all,

FYI I'll release 2.0 preview 5 on october 1st if possible.  If not, probably a few days later.

I'm not sure exactly when 2.0 final will come out. I don't want to leave it too late in the year.. but we'll probably keep finding more things we want to fix or include. There might need to be a feature freeze.

On October 1st our OSS Claude Max plan ends, and I expect to be reducing AI assistance for a bit. I was hoping to do a NoAIvember month, but it's not ideal timing. I'll see how it goes.

I think the absolute latest date I'd want to release 2.0 final would be december 19th.

A 2.0 release deserves some planning. What are your thoughts for things that would be good to have or do, beyond the usual release process ?

Best!

[Chunhui Ouyang]

Perhaps we should implement a feature freeze phase; although the PTA
likely won't have a vast number of functions, this represents a
standardized operational practice—plus, it helps prevent the urge to add
features at the last minute.

--
Chunhui Ouyang
www.qhjack.top

> --
> You received this message because you are subscribed to the Google
> Groups "hledger" group.
> To unsubscribe from this group and stop receiving emails from it, send
> an email to hledger+u...@googlegroups.com
> <mailto:hledger+u...@googlegroups.com>.
> To view this discussion visit https://groups.google.com/d/msgid/hledger/
> c770b542-cb1f-47da-9fe5-0b8bdf6a4cbb%40app.fastmail.com <https://
> groups.google.com/d/msgid/hledger/c770b542-
> cb1f-47da-9fe5-0b8bdf6a4cbb%40app.fastmail.com?
> utm_medium=email&utm_source=footer>.

[Arthur Cinader]

I am using the preview release. I implemented lots in my brokerage account, and it all foot. Awesome!

Adding each security as a commodity, the precision of crypto, lots, short lots, and the added account depth needed to support it all highlighted some shortcomings in the current hledger-web. For example, hledger-web should display lots. (https://github.com/hledgerorg/hledger/issues/2748) and support -V and -B at a minimum.

Here's what I'd like for 2.0:

Step 1: Put in the machinery for hledger-web to support the full CLI and TUI options, lots, possibly a transaction view page, and a stats page. 

Step 2: Do a first pass at organizing the new navigation elements. Plain text, existing vendored libraries - nothing fancy, but a usable set.

---- Extra Credit: ----

A. Make columns click sortable (Using query parameters and full page reload)

B. Replace existing chart graphing software with something modern. Make an example chart gallery. Probably all server-generated, but requires study

C. More thorough navigation and design scrub - replace Bootstrap

A & B could be explored independently.

All of the extra credit could be done after release.

I am working on Step 1. I pushed my WIP, which builds off Preview 5, if you'd like to take it for a test drive.

It has problems, and it isn't complete, but it is good to play with now. I don't want to rush it.

https://github.com/acinader/hledger/commit/c9606bcae

As for timing, I'm closing my September books now; I will then process 2025 tax returns as filed (extensions). That should give me plenty of real-world use with hledger-web and lots. I know there are problems with my current WIP, and I am working through them.  

If you'd like me to get steps 1 & 2 in, I'd like to target 11/1 for Preview Release 6 and 12/1 for 2.0.

Here's a screenshot of my WIP branch.

[Arthur Cinader]

updated link to my WIP: https://github.com/hledgerorg/hledger/commit/17fc2499a487d990e93792ff5752597aa69e9349

[Henning Thielemann]

On Mon, 5 Oct 2026, Arthur Cinader wrote:

> ---- Extra Credit: ----
>
> A. Make columns click sortable (Using query parameters and full page reload)

> B. Replace existing chart graphing software with something modern. Make an example chart gallery. Probably all
> server-generated, but requires study

A page that can be viewed without JavaScript would be nice.

> Here's a screenshot of my WIP branch.
>

> web-controls-2768.png

Nice ideas, though I would omit "depth" because that one can be done
easily with the query field.

I have already added toggle list mode / tree mode with and without full
paths and with and without elision and selection of layout
bare/tall/wide/barewide as a draft here. I would also like to add download
links for FODS and CSV.

However, I have a hard time working on the AI generated code. AI has no
problem to verbosely write duplicate code, redundant comments, add
translations for every combination of reports and periods, solve things in
strange ways, like building up parameter lists from scratch for every
report option switch, instead of removing the single conflicting
parameter, add tests that enforce the outputs that it found appropriate.
I'd prefer to implement in a more principled less verbose way.

[Arthur Cinader]

On Tuesday, October 6, 2026 at 2:24:56 AM UTC-7 Henning-Thielemann wrote:

Nice ideas, though I would omit "depth" because that one can be done
easily with the query field. 

I agree that the url, depth should stay in the search query and appear in the search box, but it is nice to have a search control too.

Your period/range PR made me think that it would also be nice to have a date range selector.

The two advantages of having controls on the page for all of the reporting options: 

1. It serves as living documentation of the options and query syntax. 

2. It helps a new-to-hledger, non-accountant understand how to use hledger and why the different options exist.

 

[Simon Michael]

On Mon, Oct 5, 2026, at 23:24, Henning Thielemann wrote:
> However, I have a hard time working on the AI generated code. AI has no
> problem to verbosely write duplicate code, redundant comments, add
> translations for every combination of reports and periods, solve things in
> strange ways, like building up parameter lists from scratch for every
> report option switch, instead of removing the single conflicting
> parameter, add tests that enforce the outputs that it found appropriate.
> I'd prefer to implement in a more principled less verbose way.

Could you point me to this code ? In general I haven't noticed low quality code being committed.

[Simon Michael]

On Tue, Oct 6, 2026, at 08:21, 'Simon Michael' via hledger wrote:
> Could you point me to this code ? In general I haven't noticed low
> quality code being committed.

PS perhaps in a new thread.

[Henning Thielemann]

I have pointed to some issues in my pull request:
https://github.com/hledgerorg/hledger/pull/2768

I would perform some more refactoring but would like to make sure it is
not made obsolete by more AI-generated pull requests.

[Simon Michael]

On Mon, Oct 5, 2026, at 15:06, Arthur Cinader wrote:

I am using the preview release. I implemented lots in my brokerage account, and it all foot. Awesome!

Excellent!

(You used that word "foot" again and I'm having trouble parsing it, but I think I infer the meaning!)

Adding each security as a commodity, the precision of crypto, lots, short lots, and the added account depth needed to support it all highlighted some shortcomings in the current hledger-web. For example, hledger-web should display lots. (https://github.com/hledgerorg/hledger/issues/2748) and support -V and -B at a minimum.

Here's what I'd like for 2.0:

Step 1: Put in the machinery for hledger-web to support the full CLI and TUI options, lots, possibly a transaction view page, and a stats page. 

Step 2: Do a first pass at organizing the new navigation elements. Plain text, existing vendored libraries - nothing fancy, but a usable set.

---- Extra Credit: ----

A. Make columns click sortable (Using query parameters and full page reload)

B. Replace existing chart graphing software with something modern. Make an example chart gallery. Probably all server-generated, but requires study

C. More thorough navigation and design scrub - replace Bootstrap

A & B could be explored independently.

All of the extra credit could be done after release.

Yes. We won't be "finished" for 2.0 and it'll keep evolving. And there'll be bugfix releases needed no doubt.

But I'm wary of cramming too much change just before release, without much time to discover hidden bugs and bumps through real world use.

I am working on Step 1. I pushed my WIP, which builds off Preview 5, if you'd like to take it for a test drive.

It has problems, and it isn't complete, but it is good to play with now. I don't want to rush it.

https://github.com/acinader/hledger/commit/c9606bcae

I'll do that. The screenshot is funny! And cool, maybe it could work as a toggleable control panel for power users ?

As for timing, I'm closing my September books now; I will then process 2025 tax returns as filed (extensions). That should give me plenty of real-world use with hledger-web and lots.

Excellent.

For our interest, do you use (or intend to) hledger-web pretty much exclusively, instead of the CLI ?

If you'd like me to get steps 1 & 2 in, I'd like to target 11/1 for Preview Release 6 and 12/1 for 2.0.

I was thinking the same. One more preview release on Nov 1. And aiming for 2.0 on Dec 1 (with the following week or two available as leeway in case of emergency). It probably means rather limited new features, if any, after Nov 1.

[Arthur Cinader]

On Tue, Oct 6, 2026 at 11:29 PM 'Simon Michael' via hledger <hle...@googlegroups.com> wrote:

On Mon, Oct 5, 2026, at 15:06, Arthur Cinader wrote:

...

(You used that word "foot" again and I'm having trouble parsing it, but I think I infer the meaning!)

Literally, to sum a column of numbers. In an accounting context it foots if: Assets = Liabilities + Equity. In general: "it adds up". 

In this case. I meant that it matches my brokerage account.  

...  I'm wary of cramming too much change just before release, without much time to discover hidden bugs and bumps through real world use.

Very reasonable. I thought that all I wanted was to add a balance sheet and income statement, but once I started recording lots and commodities, hledger-web became unusable without me learning about -V & -B, and depth.

I am working on Step 1. I pushed my WIP, which builds off Preview 5, if you'd like to take it for a test drive.

It has problems, and it isn't complete, but it is good to play with now. I don't want to rush it.

https://github.com/acinader/hledger/commit/c9606bcae

I'll do that. The screenshot is funny! And cool, maybe it could work as a toggleable control panel for power users ?

Awesome, I'd appreciate that.

I'll put together a draft PR so I can take feedback and iterate on it. May take a few days to open. I have already made some important fixes since that commit.

Yes, it looks funny. I have made no attempt to rationalize presentation of the "switches" yet.

BUT, not just for power users. Seeing all the switches and knobs available on the page is teaching me how to use hledger -- and how accounting works. But yes, something along the lines of "toggleable" controls for sure and almost certainly a reduced set for release.

....

 

For our interest, do you use (or intend to) hledger-web pretty much exclusively, instead of the CLI ?

Yes. I use hledger-web for (almost?) all viewing tasks. I do use the CLI occasionally. 

I use codex to collect my statements and receipts, categorize transactions, and it uses hledger to book transactions. 

I check it all using hledger-web.

 

[Arthur Cinader]

https://github.com/hledgerorg/hledger/pull/2771. 

Note the caveats and to-dos in the pr comment. Please give it a try. 

I closed #2770 because #2771 fully covers it.

I am going to leave 2771 alone for now (pending feedback) and move on to addressing issue https://github.com/hledgerorg/hledger/issues/2748, adding lots to hledger-web.