AT-LARGE GATEWAY
At-Large Regional Policy Engagement Program (ARPEP)
At-Large Review Implementation Plan Development
Page History
...
Info |
---|
Jump to the Background for the history and background of this issue. |
Table of Contents |
---|
...
Bug Reports/Observations with the new (completely rewritten) translation engine
...
December 2019
...
Below are some of the bug reports/observations noted during testing.
...
1
Missing emails from one list to another
In 2018, it was noticed that many emails are being silently dropped from one list, mainly emails from lacralo-es not being sent to lacralo-en
...
ICANN IT embarked on a complete rewrite of the translation tool in August 2019 and is ready to deploy the new tool around December 9 2019, unless showstopper bugs are identified.
The new tool is deployed on the testing mailing lists new-transbot-en and new-transbot-es
new-transbot-EN : http://mm
...
...
...
...
...
new-transbot-ES : http://mm.icann.org/pipermail/
...
...
Description of Issue | Noted by | Date Added | Status | Additional Notes on testing, fixes | |
---|---|---|---|---|---|
Bug Reports/Observations with the new translation engine (updated January 2019)
Below are some of the bug reports/observations noted during testing.
Description of Issue | Noted by | Date Added | Status | Additional Notes on testing, fixes | |||||||
---|---|---|---|---|---|---|---|---|---|---|---|
1 | |||||||||||
Month/Year | Number of lacralo-en emails | Number of lacralo-es emails | Missing emails | ||||||||
Jan 2019 | 65 | 92 | 27 | ||||||||
Dec 2018 | 63 | 71 | 8 | ||||||||
Nov 2018 | 67 | 61 | 6 | ||||||||
Oct 2018 | 63 | 70 | 7 | ||||||||
Sept 2018 | 61 | 54 | 7 | ||||||||
Aug 2018 | 76 | 69 | 7 | Jul 2018 |
Excellentable | ||
---|---|---|
| ||
Month/Year | Number of lacralo-en emails | Number of lacralo-es emails | Missing emails |
---|---|---|---|
Jan 2019 | 65 | 92 | 27 |
Dec 2018 | 63 | 71 | 8 |
Nov 2018 | 67 | 61 | 6 |
Oct 2018 | 63 | 70 | 7 |
Sept 2018 | 61 | 54 | 7 |
Aug 2018 | 76 | 69 | 7 |
Jul 2018 |
111 | 120 | 9 | |
Jun 2018 | 66 | 77 | 11 |
May 2018 | 179 | 229 | 50 |
Apr 2018 | 117 | 148 | 31 |
Mar 2018 | 56 | 80 | 24 |
Feb 2018 | 61 | 66 | 5 |
Jan 2018 | 83 | 104 | 21 |
Better identification of
- which email gives the transbot problems,
- what and where in the email gives the transbot problems
in its error response emails.
When the translation tool has issues with the email, an email is sent from transbot-no-reply@icann.org with
the static subject line "Unable to translate your email to ICANN lists" and a template text like
Dear <sender>
Thank you for your participation in the ICANN email list new-transbot-en.
You are getting this email because we were unable to translate your post automatically.
It violated one or more of the formatting rules we must impose to make translation possible.
A complete description of the formatting rules is available at:
https://community.icann.org/x/aYtEAg
In preparing your post for translation, we found the following format violations in your message:
<issue, usually Sentence punctuation must be followed by a space>
Please edit your post and send it again.
Thank you.
The format violations message doesn't say WHERE the error is in the email. If its a long email, then how can users identify and correct such issues. Persons getting these messages and not easily seeing where the problem is aren't likely to understand how to do future emails better and the warning messages becomes more of an hindrance.
Perhaps a workaround is to have the text of the email in the error email and identify what section the transbot has issues with.
Other questions :
- Why is the test required? Are the reasons for this test still valid ( translation tool can only send a limited amount of characters to the Google API)
- How are domain names handled? Since domain names can't have spaces, do domain names without beginning with http:// trigger the error?
updated
Status | ||||
---|---|---|---|---|
|
Tool now identifies the subject line of the problem email in the body of the email
<DNT> tag isn't case sensitive.
Using <DNT> tag works to not translate text, but <dnt> does not.
The tool should be able to treat the DNT tag, regardless of case.
Status | ||||
---|---|---|---|---|
|
The <DNT> tag will be seen in the original email but not in the translated one.
See EN and ES messages.
Q: Should the <DNT> tags be removed in the original email? I'm thinking it should
Status | ||||
---|---|---|---|---|
|
To achieve this, it would be necessary to re-architecture the transbot code.
Upon investigating the issue it was discovered this is a known limitation, rather than a bug in the existing code.
The request has been recorded as something to consider in a future iteration of the translation tool.
See EN and ES messages
Re: attachments, new-transbot lists have a message and attachment limit of 200K
Given many PDFs will be larger, will be hard to test unless message size limit is raised
Tested with a smaller attachment, the attachment does go through. See EN and ES
Status | ||||
---|---|---|---|---|
|
The new-transbot list email size limit has been increased to 400K.
This is enforced for the entire email, including text and attachment.
Handling an email sent to both new-transbot-en and new-transbot-es lists at the same time.
When such an email to both lists happens, some emails don't get translated.
...
- The lack of punctuation was identified as a key issue for the poor translation of emails. This is because the translate tool can only send a certain amount of characters to the Google Translate API. Without punctuation, the translation tool would have to send text mid sentence.
One of the outcomes from the LACRALO translation WG was the Proposed Notice when email is not translated message which would be sent to the user if the email had formatting issues.Subject lines would not be translated to ensure the conversation thread would not be lost and reduce the chance of garbled subject lines.
One of the outcomes from the LACRALO translation WG was the Proposed Notice when email is not translated message which would be sent to the user if the email had formatting issues. - Subject lines would not be translated to ensure the conversation thread would not be lost and reduce the chance of garbled subject lines.
FY17 update
The TTF filed a budget request to the At-Large FBSC in FY17 for ICANN to finance the hiring of a programmer to assist the volunteer ICANN staff member to fix outstanding bugs - see At-Large FY17 Budget Development Workspace , this was approved by the At-Large FBSC and filed with ICANN Finance. On the 2016-08-08 At-Large Technology Taskforce Call, ICANN Staff member Corinna Ace confirmed that a programmer/developer has been hired to sort out the remaining bugs.
To test the new translation tool, two test email lists: new-transbot-en and new-transbot-es were created and TTF volunteers and ICANN staff joined these lists to test the translation and to report bugs at discussion-of-LACRALO-mailing-list-issues page.
New versions of the translation tool were deployed to these transbot lists in late Dec 2016 and March 2017. The March 2017 update introduced new features
- translated emails will also include attachments (TXT, PDF, WORD, JPEG, PPT, PNG, GIF) from the original email
- If there is text that you don't want translated, you can enclose such text with a <DNT></DNT> tags
Since ICANN59, the TTF chairs have been discussing with Mark Segall and Corinna Ace from ICANN IT and with Silvia Vivanco and Mario Aleman from ICANN At-Large Staff on implementing the new version of the translation tool developed by ICANN IT on the existing LACRALO mailing lists.
To minimize the issue of persons posting to both lists at the same time which would create problems, members of LACRALO will be asked via online survey at https://goo.gl/forms/sEOEWqacRYLPk2Xc2 to indicate
* which lac discuss list do you wish to RECEIVE emails from (English, Spanish, or both)
* which lac discuss list do you wish to be able to SEND emails to. You can post to one list.
A conference call for LACRALO members was held on Tuesday Sept 5 2017 (see recordings at https://community.icann.org/x/yh8hB) to raise awareness of the planned changes to the translation tool used for the LACRALO mailing lists and what persons on the LACRALO lists need to do to prepare for the changes.
The tool was deployed to the main LACRALO lists on October 6 2017.
...
Testers of new translation tool for LACRALO mailing lists
...