XLIFF doesn't work. It crashes the software. I have tried every solution and nothing has worked.

[FernandoP�ramo]

So, here at the company I work, we have been using FrameMaker for decades. All of your legacy and new products are developed together with a PDF manual created in Adobe FrameMaker. This are long documents (200+ pages), and I have been using all the best practices in my documents. All my Paragraph, Character and Table catalogs are clean and lean. I'm using very litte variables and very little cross reference formats. You can see the template I'm using for the manuals I create.

Now, all of your manuals require translations into other languages. The way we work is that we send the FrameMaker files to an external agency and they handle the translations. This is extremely time consuming and expensive.

We are exploring other options, and we would like to use DeepL to handle the translations (machine learning) but DeepL requires XLIFF files. We can also try to use other tools with better DeepL integration instead of FrameMaker, but the work would be monumental to just ditch FrameMaker altogether.

I'm using version 17.0.5.725.

I have tried the solutions gives by @andreas_5547 in this thread and I can replicate the 1 kB > 74 kB workaround that he describes, but it doesn't do anything. The generation of the XLIFF files is crucial for me. It's and absolute must, and I can't get it to work.

When I try to export the XLIFF files, FM crashes without displaying any kind of error messages. It just closes. No error log is generated.

I desperetatly need help, and I can share some of my working files if necessary. I can also try beta versions of the software or experimental patches or anything. I just need this to work.

Any solutions? Any ideas?

[Jeff_Coatsworth]

Did you create brand new settings as described in the thread?

[FernandoP�ramo]

Yes, I did, and I was able to replicate the file size of the .xts file: Default (1 kB) > New settings (74 kB).

But it didn't help.

I have been trying to export XLIFF files of all the book, of a single chapter, of a single page. I have changed the font of my template, I removed all the images, I have removed all the tables. I have tried all different variations of settings and nothing works.

[Jeff_Coatsworth]

Contact the FM folks - see https://helpx.adobe.com/contact/enterprise-support.other.html#framemaker for your Adobe Support options. I'd recommend using the tcssup@adobe.com e-mail address as it reaches a team dedicated to Technical Communication Suite products including FrameMaker.

[FernandoP�ramo]

Thanks, will do.

[FrameMaker-dk]

Might it be a network issue? If your links are distributed on networked drives, that might be a possible culprit. In that case, it might be worth the while collecting the framemaker job locally with all links before doing the the XLIFF export. Frameexpert.com has a great utility for that purpose called ArchiveES.

If your book file has multiple chapters, have you tried excluding some of the chapters to see whether it is a single chapter provoking the crash?

[FernandoP�ramo]

Not a network issue. All the images are local.

I also have tried exporting the whole book, then a single chapter, a single page and so on.

[FrameMaker-dk]

If you can transfer a sample to me that has the error, I shall be glad to try and run it my end. 

info@framemaker.dk

[FernandoP�ramo]

Sure.

[FrameMaker-dk]

Thx. 

[Winfried Reng]

Hi,

Just a comment to your plan to replace your translation agency with machine translation.

I would really not recommend this!

My own experience:

o Sometimes a "not" was deleted.

o "Port" was not translated with the correct term for computer port, but with the translation of "haven" or "harbor".

o The abbreviation "cat." for "category" was translated with the term for the animal cat.

You will not want to get this in your translation!

Most translation agencies use a translation memory system which reuses exisiting translations. You can discuss, if these old translations should be checked again (and paid). Only the first translation will cost the regular fee, but updates are much cheaper.

My translation agency makes a pretranslation with these existing translations and then another round with machine translation and this gets checked by a regular translator. This reduces the costs by about 20 %.

If you would ask a local colleague to do this, you can be sure that he/she will not find all mistakes. They are not trained to do this and do not have the proper software and they might know only the target language, but not the source language.

I would discuss this with your translation agency. If you still have questions, just ask here again.

Therefore, yes, translation costs money, but this is deserved.

Best regards, Winfried

[Mats375122343lhy]

I worked as a translation manager for 20 years and can say that Winifred is spot on.

The new edition of the document is analyzed against the translation memory - which all translation projects generate if you use a commercial translation agency - and you get a result in the form of "fuzzy matches" (along a percentage axis) and 100% matches. Since the segmentation is done on a sub-para level, you typically pay a small fee for 100% matches, as they can have a new context. Note though that even the smallest change in a text segment, such as a space or a comma, will result in fuzziness. I recommend that you ask for an Excel analysis of your content against the translation memory to see the result.

Machine translation is an exciting area, but not yet at a quality level that doesn't need post-editing.

Mats

[FrameMaker-dk · Accepted solution]

For those coming here in the future with the same question, this is what can be done. We found a workaround. In this instance it was "something" in the FrameMaker documents "document properties" that made it crash when XLIFF generation ran. We found a workaround by:

• creating a new FrameMaker document
• importing all formats, except "document properties" from the troubled document.
• manually adding page size and pagination to the new document
• Copy pasting the entire text flow from the troubled document into the new document. Essentially we made a new document with only the most basic "document properties". This workaround worked and made it possible to export to XLIFF.

[FernandoP�ramo]

Thank you very much for the help.