docx problems

I don't even really know where to begin with this.
I haven't enough information to file a bug, really, because what I
have is vague, broad.

But I have been doing business as a freelance translator for 10 years,
using only FREE/OSS, (OmegaT, OpenOffice, etc., on Debian GNU/Linux)
without any major issues.

Now, suddenly, every time a client sends me a .docx file, I get a complaint
when I return their documents that the formatting has changed (tables
are different,
fonts changed, bullets disappear, or, worse, they can't even open the document,
although it opens fine here).

It's ruining my business.
I'm losing clients, losing money, and I have a family to feed.
And I don't know what to do about it.
OpenOffice won't write to .docx, and LibreOffice messes them up.

I don't know what to do to diagnose these issues, which is why
I feel I haven't sufficient information to file a bug.
I know nothing of Java (LO is primarily java, yes?),
so I certainly can't contribute to any solution.
I'm very frustrated and concerned.
I absolutely will not use proprietary software, but I am unable to serve
my clients' needs with FLOSS at this time.
I'm at a loss for what to do here.
I suppose this is venting more than anything else, and a desperate
plea for help here.

For what it's worth, I'm using the LO 3.5.4.2 in the Debian Wheezy repos,
on AMD64 architecture.

From what I gather, even MS Word still has some problems with docx, and

LO is likely to be somewhat buggy with it for the foreseeable future.
Until MS stops "developing" it so that it can become a fixed format (and
not a moving target), or makes it a truly open standard, LO is likely
to be playing a kind of guessing catch-up.

Your best bet is probably to save the file in the older .doc format. I
believe LO supports that pretty well, and your clients with MS Word
should still be able to open it just fine.

If that doesn't work, it may be the problem is when LO reads the
docx file, not when it is saving the file. In that case you'll have to
ask your clients to save the file as a .doc before they send it to you.
Sad but true: if you tell them they need to do this because you don't
use MS Word, they'll probably moan and complain, and go elsewhere, but
if you tell them it's because you use an older version of MS Word,
they'll probably tell you it's time to upgrade, but otherwise not
complain very much at all.

I'm not sure how acceptable a solution this will be for you; sadly
we're not in an age of perfect cross-compatability yet, but the fault
really does lie with MS on this one. They are the ones that cannot
make proper standards; LO does its best to keep up, but there
really is no winning. Other than using MS Word (and probably even the
version your clients use at that), there is no guarantee that docx will
work the same on your system as on your client's.

Just the situation as I understand it; others have been here longer and
may correct me.

Paul

Hi,

and you don't have the possibility to send .pdf files to your clients?
In this case you are sure that formatting and fonts will be correct.
This seem to be very important, from my point of view, especially in the case you have to translate to languages like german, polish, french where you have a lot of special letters.

To have them all correct it is the only chance I think.

best regards
gerald

Now, suddenly, every time a client sends me a .docx file, I get a
complaint when I return their documents that the formatting has
changed (tables are different,
fonts changed, bullets disappear, or, worse, they can't even open the
document, although it opens fine here).

You can ask them for DOC file, LO should handle it better.
If they are on relatively new MS Office (2007 SP2 upwards), you can ask
them to save file in ODT format. Although MSO's ODT support is
handicapped, it might be better than LO's DOCX support.

It's ruining my business.
I'm losing clients, losing money, and I have a family to feed.
And I don't know what to do about it.
OpenOffice won't write to .docx, and LibreOffice messes them up.

I hate to be the one that says it, but:
you must choose between your business and your ideas.

Microsoft has made every thing possible to retain their monopolistic
position when it comes to document exchange. They lobbied for OOXML to
become international standard, although it is underdocumented and
ill-written (XML tags are not human readable, which makes reverse
engineering harder). They screwed up implementation of their
own standard, which causes three version using it (MSO 2007, 2010 and
2013) to be incompatible at minor things.
LO is constantly getting better in terms of support of that format, but
it still has problems.

If you want to be 100% compatible with DOCX, you have to use Microsoft
Office. Period.
You may try Office 365, which is browser-based solution (if your
clients are OK with sending their documents to third party, that
is...). You will have to pay for it annually, but at least you won't
have to buy Windows.

I don't know what to do to diagnose these issues, which is why
I feel I haven't sufficient information to file a bug.

You should:
- prepare the simplest case possible, that is simplest document (even
  if it has to contain one word) that will show the issue.
- provide that document AND reference (image/PDF), so developers
  without MS Office may check if this is fixed.

I doubt this is even possible without MS Office...

I know nothing of Java (LO is primarily java, yes?),

Although it's largely irrelevant here, no.
LO is primarily C++ with some components in Java. these components are
gradually rewritten to python.

For what it's worth, I'm using the LO 3.5.4.2 in the Debian Wheezy
repos, on AMD64 architecture.

I doubt it will fix your issues (it might some of them, but perhaps not
all), but you may try to update your suite. 3.5 is quite outdated and
there were many fixes in OOXML support in later releases.

But don't even bother with wheezy-backports, as version there is 4.1.0.
It is not only outdated, (we are at 4.1.2 no, 4.1.2 should be released
next week), but also intended for early adopters. I would not run my
business on such version.

Just download deb packages from TDF website. They will install
in /opt/, and will not mess with any of your system files. You can
safely install and run them.
You can install both currently supported lines - 4.0 and 4.1 -
alongside each other without any hassle. Then you might use 4.0.6
(latest bugfix release of 4.0, intended for conservative and corporate
users) for most of work and fire up 4.1 only to see if they provide
better support for DOCX file you have received.

Switching versions with one user profile may raise some unexpected
consequences, so it is better to keep them separate. You can fire up
4.1 like that (from command line):
/opt/libreoffice4.1/program/soffice -env:UserInstallation=file:///tmp/lo-4.1-user/
This will create custom user profile directory in /tmp/, which will be
removed during next boot of machine.
If you want to preserve that directory, just change path at end of command line.
But note that there are three slashes after "file:" - two are actually part of protocol
("file://", just like "http://" and "ftp://"), third one is beginning of absolute path
at your filesystem.

I would suggest a couple things:

a) get professional support for bugs that you need fixed immediately. Since your business depends on this, I strongly suggest getting the support you need. Another option is to fo course a) fix it yourself - the code is ther (libre), or b) find a developer with free time that can specifically fix the issues you are facing.

All in all when you're using the product to make money I almost always suggest paying for professional support but first you really have to identify the issues. We need test documents and steps to reproduce. Without these things, honestly nothing we in the community can do to help.

I hope this doesn't come off as incredibly rude, just trying to say that to get anywhere at all, we need some things from your side - simply saying "it's ruining my business" is not helping but if you just want to vent - of course the community is here to listen :wink:

All the best,
Joel

Hi,

  and you don't have the possibility to send .pdf files to your clients?
  In this case you are sure that formatting and fonts will be correct.
  This seem to be very important, from my point of view, especially in the
  case you have to translate to languages like german, polish, french
  where you have a lot of special letters.

  To have them all correct it is the only chance I think.

best regards
gerald

Den 26.10.2013 00:22, skreiv baldwin linguas:

Now, suddenly, every time a client sends me a .docx file, I get a complaint
when I return their documents that the formatting has changed (tables
are different,
fonts changed, bullets disappear, or, worse, they can't even open the document,
although it opens fine here).

It is well known that .docx is a problematic format. As mentioned by others in this forum, I too asks my clients to use the .doc format instead. Explanation: If using .docx I have to use exactly the same version of MS Office that my clients are using and it is not realistic for me to buy all versions of this MS program used by different clients. This is understood, and we are happily changing documents in doc. format without problems.
BTW, my printer's refuses all docx documents.
Kolbjoern

In theory sending a PDF works, however, I have found that many files created
with LibreOffice and/or pdflatex print differently depending on the printer used.
My guess is that it is a font issue because not only is the spacing off, but the
output looks like the worst font in world was substituted.

Note: Checking embed standard fonts in LO does not solve the problem.

Best regards,

David

Perhaps you should compare the editing process using m$. Explain to a
customer that you are going to send two versions of the document (the
m$ edit version with name file1.docx and the LO edit version
file2.docx; don't tell that LO is being used!) and ask for
confirmation which version is received in better condition.

Hopefully both versions will have changed, in which case you now have
the opportunity to demonstrate to the customer that m$docx is a
dubious format to use.

If format loss never occurred with m$doc, ask the customer to send
their documents to you in that format and presumably you can continue
to use LO. However, LO is not an m$ clone and long term, you should be
advising your customer to create odf documents using LO (additional
consulting opportunity for you?)

Surely in business, you should be flexible in order to get paid? Would
you really refuse a € 1000 invoice because of the need to buy € 50
software? :wink:

Good luck.

If I sum up the problems with docx documents and with LO in general, best solution is to have a pc with linux and virtual box for running WIN7 and original MS Office.
If you earn your money with these docx then you have to adopt your work flow to that.
You can not tell your customer to change his workflow.

You will see you have a lot of problems less, doing so.

Gerald

I often work with customers who send me .docx files and expect me to edit the files and send them back.
Being aware that LO doesn't always reproduce the original formatting when saving in .docx mode, I process the files and then re-save them as "MS Word 97/2000/XP/2003". The customer's formatting is preserved and no one has ever complained about receiving a .doc file instead of a .docx.

Using LO 3.5.7.2 on Linux Mint 13 Maté.

HTH

john