Hi Erik, 2018. 06. 6, szerda keltezéssel 23.51-kor Erik Huelsmann ezt írta:

I found that, every output shows the correct UTF-8 characters, but when I enter a new value anywhere (defaults, coa, sequences) it seems the input handled as non-utf-8 and I got strange output when I open it again, like:

...

Püspökszilágy or Számla

Thanks for reporting. Is this different from 1.5.x? If not, it seems we have a problem on our hands, but it's at least not a regression.

I have no idea, I skipped the 1.5 line. I still use my 1.3.45 version in production :) Cheers, István

2018. 06. 7, csütörtök keltezéssel 22.52-kor Erik Huelsmann ezt írta:

That's perfectly fine of course. I'm wondering though: does that 1.3 version have the same UTF-8 issues? Or does UTF-8 work correctly (in contrast of the issues reported with 1.6)?

Hi, I only experienced this issue with this version: 1.6-rc2 in my local machine, running it from console: lsmb-1.6-b2/ledgersmb $ bin/ledgersmb-server.psgi  Starting Worker on PID 20268 Using Built Dojo HTTP::Server::PSGI: Accepting connections at http://0:5000/ I did not try the 1.6 in other environment or earlier versions, this is the only instance I have. István

2018. 06. 8, péntek keltezéssel 16.20-kor Erik Huelsmann ezt írta:

...

Hi, I only experienced this issue with this version: 1.6-rc2 in my local machine, running it from console: lsmb-1.6-b2/ledgersmb $ bin/ledgersmb- server.psgi  Starting Worker on PID 20268 Using Built Dojo HTTP::Server:: PSGI: Accepting connections at http://0:5000/

I did not try the 1.6 in other environment or earlier versions, this is the only instance I have.

Ok. Do you mean that this issue was tested and found not to be a problem in 1.3?

Hi Erik, Correct, with version 1.3 using apache/nginx/lighttpd this is not an issue, UTF-8 characters are working as expected.

From my point of view, this issue with 1.6-rc2:

- could related to command line start of the system - internal form processing problem (in this case it should happen with apache/nginx/etc. in background) - dojo related (?) When I enter an utf-8 text into a text box or update an existing form, it shows UTF-8 correctly, but when it saved, it will mess up and write wrong texts into the database.

It definitely happens with forms: - Tax forms description - COA - edit or add new account - description field - System/Company Address - Sequences - label It does not happen with (in other words, working as expected): - Add Part - add new part - Description, Make, Model, Bar code Bye, István

Hi István

On Fri, Jun 8, 2018 at 4:48 PM, Pongrácz István pongracz.istvan@gmail.com wrote:

2018. 6. 8, péntek keltezéssel 16.20-kor Erik Huelsmann ezt írta:

Hi, I only experienced this issue with this version: 1.6-rc2 in my local machine, running it from console: lsmb-1.6-b2/ledgersmb $ bin/ledgersmb-server.psgi Starting Worker on PID 20268 Using Built Dojo HTTP::Server::PSGI: Accepting connections at http://0:5000/

I did not try the 1.6 in other environment or earlier versions, this is the only instance I have.

Ok. Do you mean that this issue was tested and found not to be a problem in 1.3?

Hi Erik,

Correct, with version 1.3 using apache/nginx/lighttpd this is not an issue, UTF-8 characters are working as expected. From my point of view, this issue with 1.6-rc2:

• could related to command line start of the system
• internal form processing problem (in this case it should happen with

apache/nginx/etc. in background)

• dojo related (?)

Thanks for confirming.

It's a regression due to the removal of CGI::Emulate::PSGI on the majority of our code paths (which by itself is a good thing, because it gives us faster responses), but in the process UTF-8 decoding was dropped from processing FORM-tag posted values.

I've added that back and tested that creation of new customers with accented characters in the names now works and preserves the accented characters.

(And I've learned how to enter accented characters under Linux as a side-effect :-) )

Regards,