Leo 6.8.0 b1 will be delayed at least one week

Edward K. Ream

unread,

May 30, 2024, 6:43:12 AMMay 30

to leo-editor

Thomas has just discovered a serious recent error in Leo's read code. See #3957.

The fix may be straightforward, but it must be tested for at least one week.

Edward

Edward K. Ream

unread,

May 30, 2024, 7:04:30 AMMay 30

to leo-editor

On Thursday, May 30, 2024 at 5:43:12 AM UTC-5 Edward K. Ream wrote:

Thomas has just discovered a serious recent error in Leo's read code. See #3957.

The fix may be straightforward, but it must be tested for at least one week.

Still true. Happily, the error was less serious than I thought at first.

PR #3958 simply removes a do-nothing statement and relocates a comment.

Nevertheless, I think an extra week of testing is still justified.

Many thanks to Thomas for his close attention to detail.

Edward

Edward K. Ream

unread,

May 30, 2024, 7:05:19 AMMay 30

to leo-editor

On Thursday, May 30, 2024 at 6:04:30 AM UTC-5 Edward K. Ream wrote:

PR #3958 simply removes a do-nothing statement and relocates a comment.

I've merged this PR into devel.

Edward

Thomas Passin

unread,

May 30, 2024, 8:12:26 AMMay 30

to leo-editor

On Thursday, May 30, 2024 at 7:04:30 AM UTC-4 Edward K. Ream wrote:

On Thursday, May 30, 2024 at 5:43:12 AM UTC-5 Edward K. Ream wrote:

Thomas has just discovered a serious recent error in Leo's read code. See #3957.

The fix may be straightforward, but it must be tested for at least one week.

Still true. Happily, the error was less serious than I thought at first.

Good. My concern was that the statement had intended to be an assignment but by mistake did nothing.

Edward K. Ream

unread,

May 30, 2024, 8:30:04 AMMay 30

to leo-e...@googlegroups.com

On Thu, May 30, 2024 at 7:12 AM Thomas Passin wrote:

> My concern was that the statement had intended to be an assignment but by mistake did nothing.

I'm so glad you found the do-nothing statement while I still had some memory of the issues involved.

The big three checkers: pylint, ruff, and mypy failed to find the blunder. Imo, they are going backward.

pylint used to have a warning "statement seems to have no effect" but that check seems not to exist any more. I'll file bug reports to the three checkers when Leo's check_leo.py script is complete. See PR #3952.

Edward

Viktor Ransmayr

unread,

May 30, 2024, 2:33:44 PMMay 30

to leo-editor

Hello Edward,

Edward K. Ream schrieb am Donnerstag, 30. Mai 2024 um 13:04:30 UTC+2:

On Thursday, May 30, 2024 at 5:43:12 AM UTC-5 Edward K. Ream wrote:

Thomas has just discovered a serious recent error in Leo's read code. See #3957.

The fix may be straightforward, but it must be tested for at least one week.

Still true. Happily, the error was less serious than I thought at first.

PR #3958 simply removes a do-nothing statement and relocates a comment.

Nevertheless, I think an extra week of testing is still justified.

Leo's devel branch is working w/o any (immediate) issues for me in a Debian 12 & Fedora 39 VM !

From my POV the advantage of this beta would have been to also get an early feedback on the Leo @ PyPI status.

Are there any reason (other than the inherent effort of creating it), why you are concerned of doing a pre-release to PyPI earlier ?

With kind regards,

Viktor

Edward K. Ream

unread,

May 30, 2024, 3:13:39 PMMay 30

to leo-e...@googlegroups.com

On Thu, May 30, 2024 at 1:33 PM Viktor Ransmayr wrote:

>> Nevertheless, I think an extra week of testing is still justified.

Leo's devel branch is working w/o any (immediate) issues for me in a Debian 12 & Fedora 39 VM !

From my POV the advantage of this beta would have been to also get an early feedback on the Leo @ PyPI status.

That's still the purpose of the b1 release.

Are there any reason (other than the inherent effort of creating it), why you are concerned of doing a pre-release to PyPI earlier ?

I've already done most of the release-related work.

My concern is simple: I don't want to release anything until we all test the "devel" branch more thoroughly.

Edward

Thomas Passin

unread,

May 30, 2024, 4:02:33 PMMay 30

to leo-editor

On Thursday, May 30, 2024 at 3:13:39 PM UTC-4 Edward K. Ream wrote:

My concern is simple: I don't want to release anything until we all test the "devel" branch more thoroughly.

Don't we want to be testing the 6.8.0 branch at this time, instead of the devel branch?

Edward K. Ream

unread,

May 30, 2024, 6:07:43 PMMay 30

to leo-e...@googlegroups.com

Both branches are equivalent as far as Leo's code goes.

Edward

Thomas Passin

unread,

May 31, 2024, 10:05:45 AMMay 31

to leo-editor

I've just had an opportunity to check requirements.txt. One of my Linux VMs just upgraded to Python3.12 from 3.11 so didn't have any packages except what comes with a clean install. I used pip to install the dependencies from the devel branch requirements.txt and after that Leo started and runs.

Edward K. Ream

unread,

May 31, 2024, 12:46:27 PMMay 31

to leo-e...@googlegroups.com

On Fri, May 31, 2024 at 9:05 AM Thomas Passin <tbp1...@gmail.com> wrote:

I've just had an opportunity to check requirements.txt. One of my Linux VMs just upgraded to Python3.12 from 3.11 so didn't have any packages except what comes with a clean install. I used pip to install the dependencies from the devel branch requirements.txt and after that Leo started and runs.

Excellent. Thanks for your testing.

Edward

Edward K. Ream

unread,

Jun 2, 2024, 6:10:44 AMJun 2

to leo-editor

On Thursday, May 30, 2024 at 2:13:39 PM UTC-5 Edward K. Ream wrote:

> My concern is simple: I don't want to release anything until we all test the "devel" branch more thoroughly.

To elaborate: any external release (alpha, beta, whatever) that contains a serious bug will cause long-term problems that later releases won't completely fix. It's worth any amount of work to ensure that every release contains no embarrassing bug.

This principle doesn't apply to GitHub branches. We assume that updating the "devel" branch suffices to correct any blunder, no matter how large.

Edward

Edward K. Ream

unread,

Jun 3, 2024, 6:03:09 AMJun 3

to leo-editor

On Sunday, June 2, 2024 at 5:10:44 AM UTC-5 Edward K. Ream wrote:

To elaborate: any external release (alpha, beta, whatever) that contains a serious bug will cause long-term problems that later releases won't completely fix. It's worth any amount of work to ensure that every release contains no embarrassing bug.

PR #3960 continues the saga of recent changes to Leo's read code. This PR has not been merged with devel.

I'm pleased with the new code. It removes several dubious special cases and adds a message to refresh-from-disk.

I've asked the usual suspects for their reviews. I would appreciate any other review.

The new code passes every test I can think of. Please report any problems immediately.

I'm willing to release b1 this coming Friday, but I'll delay the release another week if necessary.

Edward

Reply all

Reply to author

Forward