MTEXT alignment issue 2.2.1_rc2

Version: 2.2.1_rc2-1-g4cd3b138
Compiler: GNU GCC 13.2.0
Compiled on: May 13 2024

New user, upgraded to latest rc to the try the "right click to edit Block" feature but running into MTEXT alignment issue.

We are going to fix this bug in 2.2.2.

The mtext is not easy to work with.

pclem wrote

Version: 2.2.1_rc2-1-g4cd3b138
Compiler: GNU GCC 13.2.0
Compiled on: May 13 2024

New user, upgraded to latest rc to the try the "right click to edit Block" feature but running into MTEXT alignment issue.

Why 2.2.2?
This is what release candidates are for, find errors and fix them for the final release.
MTEXT alignment works in 2.2.0, so it was broken with new features or refactoring.
Bisecting should locate the issue easy. In case it is not easy to fix, it needs to be reverted.
Just my 2 cent.

Armin

Part of my reasoning is actually already being mentioned by you.

Touching mtext code is never safe;
The fixing for mtext includes new feature: right-to-left;
We cannot simply revert some fixes: exposing a more serious bug, or breaking more features;
This issue is not serious enough for me to take immediate actions, user can workaround the issue by playing with text or alignment.

LordOfBikes wrote

Why 2.2.2?
This is what release candidates are for, find errors and fix them for the final release.
MTEXT alignment works in 2.2.0, so it was broken with new features or refactoring.
Bisecting should locate the issue easy. In case it is not easy to fix, it needs to be reverted.
Just my 2 cent.

Armin

Sorry, but I can't agree.
I'd rate a bug as certainly serious when it affects more than 90% of the users.
I have no proof for the >90%, just a guess and this is no rating or devaluing of right-to-left writing folks.

The bug, as I understand, breaks MTEXT even for existing drawings.
This will raise a bunch of issues with the new release, not only for us, but also for Linux distribution maintainers upgrading their repos.
Users request changes to reduce 4-click commands to 3-clicks or less for more efficiency. What will they do when they have to fiddle around with text alignment by trial and error to get a reasonable result for printing?
And workaround drawings, created with 2.2.1, have to be fixed for 2.2.2 again. Sound serious to me too.

We reached a new quality with the 2.2.0 release and it will be disappointing when we step back again.
Maybe we have to consider a better review process then to avoid having lower-quality pull request merged without sufficient review and testing. Once merged, the bad code get out of focus and hides in the huge code base, making it hard to debug and fix later.

It's our responsibility to take care of fixing bad pull requests or even decline them, when the concept is wrong or bad.

I added a fix to master:

1. New algorithms, instead of fixing the existing;
2. Not trying to minimize the changes;
3. A single commit, easier for backporting to 2.2.1, if determined to be safe.

LordOfBikes wrote

Sorry, but I can't agree.
I'd rate a bug as certainly serious when it affects more than 90% of the users.
I have no proof for the >90%, just a guess and this is no rating or devaluing of right-to-left writing folks.

The bug, as I understand, breaks MTEXT even for existing drawings.
This will raise a bunch of issues with the new release, not only for us, but also for Linux distribution maintainers upgrading their repos.
Users request changes to reduce 4-click commands to 3-clicks or less for more efficiency. What will they do when they have to fiddle around with text alignment by trial and error to get a reasonable result for printing?
And workaround drawings, created with 2.2.1, have to be fixed for 2.2.2 again. Sound serious to me too.

We reached a new quality with the 2.2.0 release and it will be disappointing when we step back again.
Maybe we have to consider a better review process then to avoid having lower-quality pull request merged without sufficient review and testing. Once merged, the bad code get out of focus and hides in the huge code base, making it hard to debug and fix later.

It's our responsibility to take care of fixing bad pull requests or even decline them, when the concept is wrong or bad.

I tried the MTEXT alignment fix and it works ok. Thank you.

Posts to this thread and a few others by Robertbozic looks like spam to me.

Many thanks for reporting @pclem.

I missed the first two posts, with the misplaced links.
His email address is listed on Stop Forum Spam, so I kicked him out now.

Armin

Is it possible to make the first post by approval only?

We don't get much traffic, so not a burden here.

LordOfBikes wrote

Many thanks for reporting @pclem.

I missed the first two posts, with the misplaced links.
His email address is listed on Stop Forum Spam, so I kicked him out now.

Armin