matecat / matecat

MateCat is an AI driven translation tool for language industry professionals. Matecat makes machine translation post-editing and project outsourcing easy.

Matecat

Matecat Official Repository. Matecat is an enterprise-level, web-based CAT tool designed to make post-editing and outsourcing easy and to provide a complete set of features to manage and monitor translation projects.

Installation

See the complete installation guide here:

http://site.matecat.com/installation-guide/

Available scripts

yarn format

Format all supported files using prettier.

yarn lint

Lint JavaScript files using eslint.

yarn test

Run tests related to files changed since last commit in watch mode, thanks to jest.

yarn coverage

Run all the tests and collect the coverage information and report it, thanks to jest.

matecat's Issues

mysql script error

Hi
I got following error. Could someone help me?

matecat@matecat:~/MateCat-0.6.7/lib/Model$ mysql -u root -p < matecat.sql
Enter password:
ERROR 1396 (HY000) at line 414: Operation CREATE USER failed for 'matecat'@'%'

Best regards!

Feature request: Option to hide approved segments in Review mode

Large jobs can involve enormous amounts of repetition. This is what CAT tools are theoretically most useful for, but reviewing large jobs in Matecat becomes tedious when unapproved segments become sparse. After approving a segment, the user often has to wait for a new view to load to access the next unapproved one, which adds up to a lot of waiting and clicking.

If only unapproved segments appeared in the segment queue, the wait between finding unapproved segments would be eliminated, which can speed up reviews in large jobs immensely(!).

In terms of implementation, the server could be asked to send only unapproved segments, and the client could assemble them into a temporary view.

Analysis issue

Hi there,

Thanks for your fixing with Revise function.

I am checking the system again and it notifies me with a new bug.

It seems Analysis function didn't work in somewhere.

1/ In Volume Analysis page, it always display as Fast word counting... Also it doesn't breakdown to fuzzy match table.
2/ In Translate/Revise editor window, it always display as Calculating word count...

I attach here screenshot for your reference.

Thank you very much!

Kevin

Error when testing in virtual machine

I download the VirtualBox 5.0 version, and import to my machine. Everything going well.
But when start to test for translation, I got 2 strange error.

1. "Fast word counting.." never stop, and the Split button never turn enable.
   
2. click the translate, open the translator page. Nothing in the page but the top bar and bottom bar with "calculating word count..."
   

The interface slows down progressively.

This is because old segments, which all hold a copy of the editor class (why?), are kept on the page as new ones are loaded. This causes the memory footprint to grow and the interface to get progressively slower, until the user reloads. I posted a video showcasing the slow performance in issue #18 : https://www.youtube.com/watch?v=yiQl2pvmcak

Keeping old segments in memory would not be an issue if it did not affect performance significantly, but currently it does. I suggest that old segments at the top of the page be discarded after a maximum number of segments has been loaded. The number can be a user-specific setting.

Including untranslated segments in Export Job TMX

Currently the Export Job TMX function does not include untranslated segments.

If untranslated segments where available in the exported Job TMX, it would allow for use cases other than QA, such as, doing concordance searches with Xbench among non-translated segments yet.

For larger projects, that helps researching terminology and glossary preparation because you can see all segments in you job where a given term will appear (you do not have to wait until the translation is finished to understand that the term chosen was wrong).

Google sign

Is sign in in Google the only way to manage when using a own server or virtual box?
And are the TMs and files saved into my Google account?

FetchALL from segment_translations based on id_job

This fetchAll could be heavy

https://github.com/matecat/MateCat/blob/develop/lib/Model/Translations/SegmentTranslationDao.php#L85

At moment this function is used only in tests.

php tag on PHP 5.6

Hi
Just installed MateCat on Debian 8.1
PHP 5.6.12-0+deb8u1 (cli)

I checked it out from git
There following 4 files are missing the starting <?php using the old <?
This is not working anymore on PHP 5.6

lib/Utils/CatUtils.php
lib/Utils/MyMemory.copyrighted.php
lib/Utils/Engines/Results/MyMemory/TMS.php
lib/Utils/Engines/Results/MyMemory/Matches.php

Thanks for an amusing software

Are Casilla

Revision field sometimes contains text from previous segment, blocks progress

As seen in these screenshots:
http://imgur.com/kwOt43g
http://imgur.com/tKnXgSJ

The text in the Revision field is not identical with the segment, so it cannot be approved. Moving away and going back to the segment usually does not fix the text, although sometimes it does. Most often, the page has to be reloaded or the segment skipped, which interrupts the workflow. When the issue is persistent, it forces me to skip 20-40% of segments, but only after having read and trying to approve them.

The text in the revision field is almost always the content of the segment before the one being displayed. In the first screenshot, the segment containing "Betriebsdruck" was autopropagated and skipped over. One other time though, I approved 5 segments quickly in a row, and the second segment's text appeared in the Revision field.

It feels like this has to do with approving segments faster than the editor pane opens/finishes loading. Approving segments slowly and waiting for the pane to load normally causes no issues. I currently have this issue a lot with ping results like this:

--- matecat.com ping statistics ---
171 packets transmitted, 170 packets received, 0.6% packet loss
round-trip min/avg/max/stddev = 269.694/320.451/460.120/39.874 ms

I'll see if those are any different when I don't notice it as much.

Why can only process XLIFF file?

I deploy MateCat in my own server with the install.txt in github.
But in the homepage, I saw "Translate an XLIFF file", but not "Translate your files". And I can only upload XLIFF file.
Is there any more config for the application?
And another question is pages such as opensource/faq in https://www.matecat.com/ return with 502 error? So that I can't get any documents of the project.

Oauth gets stuck after entering Google credentials

I installed a self-hosted Matecat following these instructions on Ubuntu 14.04: http://www.matecat.com/advanced-manual-setup/

Everything went fine except that under Enabling Google+ login I had to deviate slightly from the commands in the instructions:

chmod 400  ~/cattool/inc/oauth_config.ini
chown www-data:www-data ~/cattool/inc/oauth_config.ini

because the presumed matecat user (i.e ~ stands for /home/matecat) did not have permission to run the second command. So instead I ran this command as root:

chown www-data:www-data /home/matecat/cattool/inc/oauth_config.ini

Now, when I try to use the Google authentication, it asks and validates the Gmail credentials, but the response gets stuck (the Google Sign In pop up is completely blank and does not disappear) with this url: http://<somehostname>/oauth/response?code=<somecode>

Also the source code for the Google Sign In pop up is completely empty (no HTML code at all).

To double-check if the address /oauth/response was correct at all, I opened from a browser this url: http://<somehostname>/oauth/response (no code attribute) In this case, the page source code is not empty but this one, so I guess that the address is correct:

<html>
<head>
<script type=text/JavaScript language=javascript>
function doload(url){
        if( url != '/' ) {
            window.opener.location.href = url;
        } else {
            window.opener.location.reload();
        }
        self.close();
}
</script>
</head>
<body onload="javascript:doload(&#039;/&#039;);">
<p id=loader style="display:none; margin:130px auto 0 auto;text-align:center"><img src=http://www.translated.net/it/pub/images/loading.gif></p>
</body>
</html>

Did I do anything wrong with the workaround of using root instead of matecat to avoid the permission error when following the instructions under Enabling Google+ login?

Double spaces cannot be replaced by a single space

Because the replacement string is considered empty.

i18n support?!

Hey guys, couldn't figure out how to show translation keys when doing i18n like translations. Is there a way? Would be very helpful! Thanks

What dependencies can we update till newest versions for MateCAT 1.0?

Hello,

GOAL: I would like to upgrade all or most components so that there were no security vulnerabilities (e.g. PHP 5.0 is not supported anymore) and performance was improved (PHP 7.0 give performance boost for PHP projects).

I will upgrade MateCAT often, therefore I better stick to the versions you use and test yourselfs.

PROBLEM: after installing the below components from scratch and following the Advanced Installation Guide, the MateCAT 0.6.8 was showing empty screen (I did not try 1.0 yet) and only worked when I installed the versions from MateCAT installation guide):
• Ubuntu 16.04 LTS
• Apache 2.4.2
• PHP 7.0.5
• MySQL 5.7
• git 2.8.1
• ActiveMQ 5.13.2 (tested, it works)
• Redis 3 .0.7
• Node.js 4.4.3
• Java 8

QUESTION: tell me which LATEST versions for the above dependencies are OK to use with MateCAT 1.0 e.g.:
• Ubuntu 14.04 LTS
• Apache 2.4.2
• PHP 5.0
<....>

Thanks,
Mindaugas

Find and Replace includes segments not assigned to you, allows replacing texts in them

As described in title, Find & Replace not only includes segments that are not assigned to you, but it also shows no error message when you replace text in them.

It doesn't look like changes done this way are saved on the server, but it's unintuitive and not very helpful. This is related to #20 in that the Find & Replace function feels unresponsive.

Smooth scroll upon opening a segment recenters view at the end, even if user scrolled away during it

This is not only annoying during revision when you approve a segment, try to scroll down, and have your view ripped away, but it also seems to stack when you hammer Cmd+Enter 10-20 times to approve the next section of text, and you are unable to scroll for a while.

chmod: cannot access ‘www-data:www-data’: No such file or directory

Hi,

I managed to install MateCat in Ubuntu Server 14
Quite at the end, I'm finding some difficulties when entering:

chmod 400 www-data:www-data ~/cattool/inc/oauth_config.ini

I receive the message:

root@server:~/cattool# chmod 400 www-data:www-data ~/cattool/inc/oauth_config.ini
chmod: cannot access ‘www-data:www-data’: No such file or directory

Do you know the reason for this?

Status of segment in Export Job TMX missing

Currently there is no way to know if a segment in a exported job TMX is translated or approved.

Sometimes when you QA a translation, it is useful to know the segments status, in order to make a decision when several options are available.

I suggest to add a status metadata to the segment. For example:

<prop type="x-MateCAT-status">translated</prop>
<prop type="x-MateCAT-status">approved</prop>

Suggested feature for SDLXLIFF: segment status (75% fuzzy, 100% match, Locked)

Hello, Developers,

We would like to work on SDLXLIFF files, pretranslated in Studio 2014, then imported to MateCAT.

CONTEXT: we upload SDLXLIFF file (e.g. English-German), pre-translated in SDL Trados 2014. There are NoMatch segments, some segments are LOCKED, some FUZZY matches (but NOT from MyMemory), some 100% matches, some PerfectMatches.

PROBLEM: in MateCat all segments statuses are lost and translator does not know whether to Skip 100% matches or not, and which segments are FuzzyMatches. She needs to read every segment :(.

QUESTION: when do you plan to implement showing segment status in MateCat for XLIFF or SDLXLIFF format?

Thanks,
Mindaugas

Segments in Exported Job TMX for split jobs do not distinghish the split part they belong to

When exporting the Job TMX for a split job, currently it is not possible to tell which segment belongs to which of the split parts.

Additional metada would be needed for split jobs so that a QA process knows which is the relevant split part for the current project_key.

For example, something like this could be added for split jobs to the segment metadata:
<prop type="x-MateCAT-id_split">2</prop>

Suggestion: a bulk approve function during revision

Opening segments during revision takes a non-negligible amount of time, especially on older machines. There are sections that would be very quick to scan and review, but opening each segment and waiting for the auto-scroll to finish makes those take a disproportionately long time in Matecat.

Options to "Approve the next X segments" and "Approve segments from X to Y" would be extremely helpful, and the next best thing outside of a proper selection mechanism. A "Yes, auto-propagate all those approvals" checkmark would be good, too.

I would be glad to see somehing like this - it would at the very least save me a few hours.

Displaying engine error messages during translation

Error messages that are returned by an MT engine during translation aren't shown to the user.

This is a problem in the case of the Tilde MT engine. If a Tilde translation system has not been used for a while it enters a stand-by mode. It gets brought up again after a new translation request is made but waking up a system might take up to 10 minutes. This is a very common scenario so there really should be a way to let the user know why he isn't getting any translations from his MT engine. He might not be getting a translation due to a system still waking up or a system could have been turned off or deleted without the user knowing (by some other user). So there should be a way to let the user know whether he should just continue working while the system gets brought up or should he not expect a translation without some further actions (e.g. going to the Tilde MT site to check on the system).

Are there any plans to implement a mechanism to show error messages or some king of notifications during translation?

Feature request: Mark a comment as unresolved

Reason: there is no Undo option for when you accidentally resolve a comment.

After installing build 1.0 from scratch, localhost in browser shows empty screen

Hello,

My environment: Ubuntu 14.04.1 LTS (as 14.04 LTS is no longer supported) and ActiveMQ 5.13.2 (as the version 5.11.3 from your manual is no longer available).

CONTEXT: after upgrading working Matecat 0.6.8 installation to 1.0 the browser showed only empty screen with favicon in the browser tab. No error message. Restarting Apache or Ubuntu does not help.

I HAVE TRIED ALTERNATIVE: I also installed Matecat 1.0 on clean Ubuntu, hoping that clean install will work. No, it does not. Even after restarting Apache or Ubuntu.

PROBLEM: neither upgrading from 0.6.8 to 1.0 nor clean install of 1.0 works, the browser shows the empty screen (only Matecat favicon).

Please help to troubleshoot the cause of the empty screen.

Thanks,
Mindaugas

Adjacent spaces are eliminated when pasting text over a selection.

Let me know if you need a screencast.

In Revise mode, auto-skipping to the next unapproved segment does not work if none are in memory

Normally, approving a segment skips to the next unapproved segment. This does not work if no unapproved segment is left in the current view. Instead, the segment below the last is opened. The workaround is to click on the progress bar manually (this sometimes takes a few clicks for some reason) until the next unapproved segment is selected.

This issue is especially inconvenient in large jobs where auto-propagate fills in many segments, and remaining segments may be strewn throughout the document, requiring many skips. It forces the user to do the skipping manually and also to keep an eye on the segments' current status, or they risk re-reviewing their work and wasting time.

A quick fix would be to simply call the same function as the progress bar if no unapproved segment is found in current memory.

Find&Replace selection moves cursor when segment is edited

As seen in this video: https://www.youtube.com/watch?v=C56PgH0AKk4&feature=youtube_gdata_player

Deleting parts of the selection does not move the cursor, only adding text does.

T+>> fails if the next segment is identical to the current one

In the example arrangement below, translating segment 1 and pressing the T+>> button will jump to segment 3. However, since the translation auto-propagates from segment 1 to segment 3, T+>> opens an already translated segment.

...
Segment 1 / Content A
Segment 2 / Content B (already translated)
Segment 3 / Content A
...

Obviously this behavior is not intended and causes unnecessary button presses for the user. A solution would be to compare the content of the next segment to the last submitted one locally, and continue the search for an untranslated segment if it is identical.

A similar problem exists in Revise mode, see issue #17.

how shall we protect our MateCAT installation from DDOS attacks?

Hello,

I am sure that you took care to make PHP code secure, however, www.matecat.com is open to public.
How do you protect it from outside threats (as attack would paralyze all Translated.net).

Do you use Cloudfare services, or just a good hardware firewall?

Thanks,
Mindaugas

MySQL Data saving + Updates

Hi

So, the translation units are saved into MySQL

Is it possible to run MySQL and save its DATA in a different system using Windows?
A kind of tunneling?

Will you consider using SQLITE in the future?

The second question is: How easy it is, to update MateCat to a newer version?
Over the linux command line?

Best regards!

Alf

Tilde MT integration

I'm opening a new issue to use for communication on Tilde MT integration.

Here is a preview of basic Tilde MT integration tilde-nlp/MateCat/commits/lets-mt-integration. This basically just follows the stuff that was already there for MicrosoftHub and Moses engines.

The above code works but the user has to enter three different ID's (Client ID, System ID, Terms ID) to add our engine. So we would like to display some basic config options after the user enters the Client ID so that the System ID and Terms ID can be selected from a list. Here is a preview on how this might look tilde-nlp/MateCat/commits/letsmt-advanced-integration.

I add the possibility to specify another template to be displayed after the user presses Confirm in the initial view. The template is then filled with data that are returned by the engine (in place of the id of the newly added engine that would be returned in the usual scenario).

The stuff that is not yet implemented is some custom JavaScript that would be attached to the System selection dropdown onchange event (in the newly added engine config view). This is necessary to be able to return the available term collection ID's after the user selects a system from the dropdown. When the user clicks an element in the dropdown an AJAX call would be made to fetch the relevant term list and fill out the term dropdown.

This is the basic idea. Does this seem reasonable?

The Auto-Propagate dialogue proposes no hotkeys

It forces a mouse move and click every time it comes up and steals time during big jobs. Cmd+Enter and Escape respectively might be good hotkeys.

if we install MateCat-tagProjectionEnIt build, can it be used with MMT?

Hello,

We would like to start using MateCAT with MMT in production as soon as possible (as MateCAT itself works fine, we only need MT).

1. can we just download MateCat-tagProjectionEnIt build and install so that we could work with MMT? I noticed that MMT.php is a very fresh file.
2. can MMT be used in production? I know you are still developing MMT, but I tested with your demo data and translations are being generated fine. Can we used it in production, together with our own installation of Matecat?

If answers are negative, then when you plan to integrate MMT and Matecat so that is can be used in production.
(if Github should not be used for questions like this, only real issues, then let me know, I will use email))

Thanks,
Mindaugas

PDO Exception, matecat.sql file lack fields

Version 0.6.1e affected

Inserted spaces are not parsed correctly in the "Revision Field (Track Changes)"

Inserting a space to revert a segment to its original state causes the Revision Field to display an error, even if there is none.

Screencast:
https://www.youtube.com/watch?v=M6f-q3bov38&feature=youtube_gdata_player

Auto-Propagation does not work properly on currently displayed items, causes usability problems

I am revising a job with many identical segments grouped together, which causes some irritating and inconvenient behavior that costs a lot of time. One of the main issues is the fact that auto-propagating does not seem to affect the segments currently on screen.

For example, if I make a change to the first group of 6 identical segments and press Cmd+Enter, the Auto-Propagation dialogue pops up, but confirming it will not change the status of the other 5 segments that are visible then. Not only that, but it will also continue on to the second of the six segments and effectively prompt for the same work that has been done on the previous segment, which is at best a waste of time, at worst a source of errors.

If I approve the first in a line of identical segments, the rest will not be automatically approved or skipped, and Cmd+Enter prompts for a response to each of them. Only segments loaded from the server at the end of the current pre-loaded segments are marked as approved. When clicking on the progress bar to skip to the next new segment, the already-approved-but-not segments are usually selected 1 or 2 times before it is registered that they are not new. This might be the same or a related issue, or due to something else.

Either way, all this makes working on this kind of job extremely time-intensive and sluggish, and can trip up users if they are not careful. It would be great if this was fixed.

Feature Request: "Approve and do not go to next unapproved segment" button in Revise view

There are situations when one has to change a handful segments that are already approved, but approving them always jumps to the next unapproved field. This behavior is optional in translation mode, but enforced during revision mode. A separate button like in the other mode would be coherent and helpful.

Connection to server gets lost randomly (Review mode)

This has happened to me at least 10 times today and 3 times in the last hour. Approved segments stop turning green, and waiting for 10-30 minutes does not change their status. There is no warning or alert for this, and there seems to be no fix except reloading the page. After the reload, approving segments works fine, but the segments that were not yet transmitted are lost.

It should go without saying that losing users' work is a severe bug, and that this behavior is severely frustrating.

I have a hunch that it is connected to #20, because it seems to happen most often when I am progressing fast through the segments. That might well just be confirmation bias, however. I could not reproduce this reliably yet, but then again there is no connection information displayed in the client that would help with doing so.

While the underlying issue is being fixed, a temporary hotfix that shows an alert when more than X segments are queued for the server would be much appreciated.

Revision function doesn't work

Hi there,

I am trying version 0.5.1 in VM. Everything goes well except the Revise function. After translation, we click Revise in the drop-down menu list in Manage section, then a 404 Not Found page appear.

Would you please check?

Thanks!

Feature request: Hotkeys for Revise pane

Every time a segment is edited during revision, submitting it requires moving your hand from the keyboard to the mouse (since you just typed), clicking one or more buttons in the "Select the type of issue" field, and moving your hand back. This feels unnecessarily cumbersome and uncomfortable.

It would be great if hotkeys like this existed:

Ctrl/Cmd+1: Move the "Tag issues" radio button selection one to the right (or back to the far left if at max)
Ctrl/Cmd+2-5: The same for the other 4 error types.

These are used for tab selection in the browser, but since Ctrl+F is already overloaded, they might as well be. If changing the radio button status steals focus from the text editing field, this hotkey would be useful as well:

Ctrl/Cmd+E: Move the cursor to the editing field of the selected segment

I believe this will not inconvenience normal users, and increase usability and revision speed.

Open brackets, tags in search term cause faulty behavior

If you search for a term like "PA (Nylon", clicking on "Find" does not give a result. If a previous search has displayed a result, the display will be updated to say "Searching..." but never finish, or otherwise not pop up at all. The function seems to crash before that point.

This issue is especially relevant when tags are inside brackets, because those also seem to cause problems during search and it's not clear how to deal with them. I am trying to find a segment that reads " PA (Nylon ) " according to the QA report tool, but that term does not seem to be searched correctly either.

Approving segments in a completed review fails

When all segments in a review are completed and you try to approve a segment, the selection jumps to the start of the page. This makes it almost impossible to correct a review after the fact. See here for a video.

Solution: Split the "Approve" and "Go to next unapproved segment" functions into two buttons, like it is done for the Translate view.

As a side note regarding performance, this was done on a top of the line PC.

When Revise Pane is folded in, approving a changed segment fails silently, shows no error

Normally, trying to approve a segment after changing it will highlight the "Select the type of issue" field in red, but this is invisible when the pane is not expanded. In that case, the button will simply not react and it will look like a bug or connection error is at fault.

A solution would be to expand the field automatically upon clicking the button if further action is necessary.

In Editing Log, "Back to Translation" takes you to Translate mode even if you were in Review mode

If you are revising a project, opening the "Editing Log" and clicking on the "Back to Translation" button takes you to the Translate view of the project, instead of the Revise view that you were in originally.

Intuitively, I believe the button should lead back to the same mode that you started in.

Feature request: "NOT" keyword for searching

To check if a word has been inconsistently translated, it would be very helpful to be able to use a "NOT" keyword in the search field. Is that already implemented but not documented? How much effort would it be?

Find & Replace does not save changes reliably, seems to lose connection to server

It is unclear when Find & Replace results in a lasting change to a segment. If I progress through the matches and replace manually, I often end up running in circles, as entries I have already changed come up again as matches.

I have not tested and reproduced this reliably. It happens about half the time for me. This issue is especially treacherous because it is silent and easy to overlook that the changes have been lost. I have (embarrassingly) submitted work with mistakes that I am absolutely certain I had fixed via Find & Replace.

Intuitively, I'd guess that the changes might be discarded when a new view / new segments are loaded from the server. When are changes written to a segment / uploaded to a server? Maybe information is lost at that point?

edit: I should add that I have been using the Revise mode, if that changes anything.

Exception: Engine 1 not found

Using the 0.5.1 version, I am getting the following exception when I upload an XLIFF file and click "Analyze":

[ E_ERROR ]
INIT-> fatalErrorHandler (line:434) :   Uncaught exception 'Exception' with message 'Engine 1 not found' in /app/lib/utils/Engine.php:36
INIT-> fatalErrorHandler (line:434) : Stack trace:
INIT-> fatalErrorHandler (line:434) : #0 /app/lib/utils/TMSService.php(34): Engine::getInstance(1)
INIT-> fatalErrorHandler (line:434) : #1 /app/lib/utils/ProjectManager.php(81): TMSService->__construct()
INIT-> fatalErrorHandler (line:434) : #2 /app/lib/controller/createProjectController.php(218): ProjectManager->__construct()
INIT-> fatalErrorHandler (line:434) : #3 /app/index.php(15): createProjectController->doAction()
INIT-> fatalErrorHandler (line:434) : #4 {main}
INIT-> fatalErrorHandler (line:434) :   thrown
INIT-> fatalErrorHandler (line:434) :   Not Recoverable Error on line 36 in file /app/lib/utils/Engine.php - PHP 5.5.9-1ubuntu4.7 (Linux)
INIT-> fatalErrorHandler (line:434) :  - REQUEST URI: /?action=createProject&time=1427843506666,jid=undefined
INIT-> fatalErrorHandler (line:434) :  - REQUEST Message: Array
INIT-> fatalErrorHandler (line:434) : (
INIT-> fatalErrorHandler (line:434) :     [action] => createProject
INIT-> fatalErrorHandler (line:434) :     [time] => 1427843506666,jid=undefined
INIT-> fatalErrorHandler (line:434) :     [file_name] => xliff.xliff
INIT-> fatalErrorHandler (line:434) :     [project_name] => 
INIT-> fatalErrorHandler (line:434) :     [source_language] => en-US
INIT-> fatalErrorHandler (line:434) :     [target_language] => fr-FR
INIT-> fatalErrorHandler (line:434) :     [job_subject] => general
INIT-> fatalErrorHandler (line:434) :     [disable_tms_engine] => false
INIT-> fatalErrorHandler (line:434) :     [mt_engine] => 1
INIT-> fatalErrorHandler (line:434) :     [private_tm_key] => 
INIT-> fatalErrorHandler (line:434) :     [private_keys_list] => {"ownergroup":[],"mine":[],"anonymous":[]}
INIT-> fatalErrorHandler (line:434) :     [private_tm_user] => 
INIT-> fatalErrorHandler (line:434) :     [private_tm_pass] => 
INIT-> fatalErrorHandler (line:434) :     [lang_detect_files] => Array
fatalErrorHandler (line:434) :         (
INIT-> fatalErrorHandler (line:434) :             [xliff.xliff] => detect
INIT-> fatalErrorHandler (line:434) :         )
INIT-> fatalErrorHandler (line:434) : 
INIT-> fatalErrorHandler (line:434) : )
INIT-> fatalErrorHandler (line:434) : 
INIT-> fatalErrorHandler (line:434) : 
INIT-> fatalErrorHandler (line:434) :   Aborting...

The same file works well on the hosted version. Any idea what I am doing wrong?

Windows Server

Hi,

Is it possible to install this in a Web Server like ZPANEL using WINDOWS?

The hotkey for T+>> should always be available

The "T+>>" button is only available when the next segment is translated. When it is not, the hotkey does not work, although it should, since the user input "This is translated, take me to whatever the next segment is" still makes sense.

Currently, the user has to manually parse if the next segment is translated and select the right button/press the right hotkey to proceed to the next untranslated segment. If you mess up, you either pressed a hotkey for nothing (bad), or you are taken to a segment that has already been translated (bad). There is no reason to make the user do this.

The best solution in my opinion is to keep the T+>> button visible at all times. Sometimes its behavior will be identical to the last one, but since it will not confuse or overload the interface, that should be no problem.