Mac ASCII to Win/Linux ?

I’ve limited my Scrivener input to basic ASCII … that is, no curly quotes, ellipsis, accents, em dashes, etc. … when drafting on my Mac PPC/OS ‘Tiger’, because a page of Scrivener for Mac text will show up shot full of “&xx” replacements when opened in Scrivener for Windows, or Linux. The only way I’ve found to convert them is a laborious search and replace conversion. It’s easier to avoid extended ASCII as a result.

There seems to be no Scrivener menu control for text coding (?). Using UTF-8 coding via word processor intervention doesn’t fix it. I’m stumped. I’m aware that Mac and Win use different codes for extended ASCII, but I’d hoped that somewhere there was a middle-ground of UTF-8 translation/intervention? It seems not …

Am I stuck with basic ASCII to edit the same Scrivener project between Mac <==> Win / Linux?

I haven’t actually tested Linux here, but I’m not having any difficulties with basic extended characters like accented Latin and typographic punctuation. What fonts are you using to input, on the Mac side, and to display on the PC side? It might also help to know how you are transmitting the files. For example are you uploading them to your web server with an FTP client and downloading them with FTP on the PC? I’m sure you are not—but you get my point: a strange workflow that uses tools that could potentially alter encoding should also be a part of the examination.

My test: I turned on all fancy punctuation on my Mac and created a test project with some accented Latin and such. I closed this, switched over to the virtual machine running Windows 7 and opened the project using 1.6.1. Everything displays flawlessly. Origination font: Cochin; display font on Windows: Helvetica.

Ioa, they are on PPC (Tiger/10.6) which I think uses a much older version of scriv. Not that you didn’t test that…

I was just thinking that it might be some old bug from a long time ago that isn’t present today…

Maybe… though Tiger (10.4) should be using UTF-8 just as OS X always did, all the way back to the Cheetah (though more complete script support was not around until 10.2, but we’re talking Arabic ligatures and such here). You’re right I didn’t specifically test on a 10.4 machine though. But just to check, I have attached a .scriv project. It has a file I typed on the Mac, then duplicated. I edited the duplicate on the PC. Both files look as they would be expected to. See if that works on the PC.
special_characters.zip (38.6 KB)

Tiger was 10.4?

I must need a new SSD brain.

First off, thanks for the kind replies. Secondly, color me somewhat confused. :blush:

I’ve run a series of tests between three machines:

Mac Pismo w/G4 processor upgrade, running OS-X 10.4.11 “Tiger” & Scrivener 2.5, most recent version;

Lenovo ThinkPad T-400 running Win 7, SP1 (32bit) & Scrivener 1.6.0.0 for Windows & Linux

Compaq Presario S4000T running Makulu (Debian Testing, XFCE, 32-bit) & Scrivener for Linux 1.6.1.1

All file transfers were done with a USB flash (thumb) drive formatted FAT-32 (This is the single most ‘universal’ drive format I’ve found for transfer between Mac-Win-Linux.)


Results are somewhat inconclusive, as I did get some inconsistent results which I’ll need to redo most carefully sometime later, taking careful notes. But in general …

Mac to Win produced faithful display of extended ASCII characters, but loss of font name data, except for Times New Roman.

Mac to Linux produced loss of extended ASCII characters, with substitution of “&xx” coding, and loss of font name data.

Taking the files back to Mac retained the “&xx” substitutions imposed in Linux.

(At this point, Windows is the winner. It faithfully displayed whatever was brought to it.)

EXCEPT … going from Mac Scriv to Win Scriv to Linux Scriv … and then back to Mac Scriv … I lost the entire .RTF document display. It was corrupted and displayed the .RTF encoding on top of the original text. (So, as Groucho Marx used to say in his Doctor skits … if it hurts when you do that, then DON’T DO THAT!)


Composing a Scriv project in Linux (Debian,XFCE) with Scriv for Linux 1.6.1.1 … absolutely beautiful display and responsiveness. A joy to work with. Taking the project via USB thumb drive back to Mac Scriv 2.5 on OSX 10.4.11 … perfect display of extended ASCII, but loss of font name data.

Conclusion: ??? (What’s going on here?)

I can only assume this will require much more careful note-taking and control of variables to nail down just what is happening. However, once again I’m reminded that nothing can be taken for granted with computers, and nothing is ever as simple as it seems.


I’m a bit amused at the reaction to “Tiger” in this thread. Let’s just say that Apple charged forward entirely too furiously for this old graybeard to keep up, and I have two absolutely choice PPC machines: the Pismo laptop with G4 upgrade (the best, most versatile laptop Apple ever produced) and a PowerMac MMD G4 1.45 Ghz dual processor tower (yes, the “wind tunnel” machine!) Both are loaded with software I still cannot find equivalents for in either Linux or Windows. Nope … not gonna dump them to get something several thousand$$$ more shiney!

Thanks for the help. My basic conclusion is, for the time being, to have confidence in Tiger Mac–>Win and Win–>Tiger Mac; and Linux–>Win or Linux–>Tiger Mac … but NOT at this time, to go Tiger Mac–>Linux.

Oh … the reason for futzing with Linux? MS is killing Win XP in April, and I’ve got a pair of ruggedized salt-water resistant WinXP laptops for navigation software on my Tartan 30; and the wife has an XP machine with a couple thousand $$ worth of embroidery/sewing software. I can’t transition those to Win 7 without extreme pain, so I’ll air-gap them to isolate 'em, and dual-boot Linux for hazardous (Internet) duty. Now that has been an interesting journey …

=gb=

Ioa, thanks for that test file. I unzipped it to a USB thumb drive. Opened both test files on the Linux machine, Scriv Linux 6.1.1 – perfect! Opened both on the Win7 laptop, Scriv Win 6.0.0 – perfect! Opened both on the Mac Pismo 10.4.11 laptop, Scriv Mac 2.5 – perfect!

I did NOT try editing/writing your test file; I’m pretty sure if I modified it on the 10.4.11 Mac and then took it to Linux, it would be corrupted and display the ?d2 … etc. character substitutions. I’d even bet a pint of cold suds on that. I’m fairly sure that Mac Tiger → Win7 would be okay, tho …

Thanks for that file; it’s a nice confirmation. And as for corruption at my end, I’ll repeat again what Groucho Marx said: "If it hurts when you do that, then DON’T DO THAT!

=gb=

I empathise with the predicament. I sometimes feel quite in between platforms as well. Apple is increasingly moving toward a fixed point that I have no interest in (tablets and vendor lock-in junk), and so is Microsoft for that matter. I increasingly find myself on Debian GNU/Linux these days. Not yet for “real work”, I still have too many Mac tools for that to happen (mainly Scrivener and scattered others), but to keep my toes in that pool of water, so that if/when I have to jump in, the shock isn’t too much. As for myself I stopped at 10.6 rather than 10.4. I think it’s the best OS they made, even if it was largely an apology over the travesty that was 10.5. :slight_smile: 10.4 was a good place too, and I still admire the PPC chip. The Pismo is probably the last true workhorse laptop Apple made, but I will always be fond of the original Titanium G4, which is undoubtedly one of the most influential hardware designs. That thing was technological art from the top down, though fragile as a butterfly. :slight_smile:

Anyway! Keep that thing running, you probably won’t ever see anything like that from Apple again.

Yes, FAT32 is the best common denominator in my experience as well. Everything else requires somewhat clunky drivers, and I get the willies when it comes to creaky drivers messing with a filesystem.

I will summon Jennifer for this one. There are a few things which are familiar in passing. In particular the condition where raw RTF code ends up in the editor is something she is working on right now with Chinese and English text. There may be some commonalities there that could aide the bug fixing effort on that score. It could be whatever triggers it from outside of the UTF-8 ASCII compatibility zone, be it Chinese or a curly quote, is related.

Here are a couple of things to consider. Not answers, alas.

  1. Fonts will always depend on the fonts available to the host OS, yes? Editing a doc will usually substitute locally- available fonts for whatever fonts were specified on the other OS. So it is probably best practice when hopping from one box and OS to another to keep only one OS as the compile OS, with font overrides in its compile settings.

  2. Upper-ASCII in *.rtf is specified not at utf-8 but in control sequences starting with back-slashes as in TeX and Troff, as I recall. There have been a few different versions of RTF coming from Microsoft, but I don’t know whether the accented character control sequences changed along the way. Is it possible that there are versions of RTF that use utf-8 instead of the control sequences, and others that use the control sequences, and the corruption is coming in trying to translate them?

  3. Linux' != Linux’ as different distributions of the collection of programs we call `Linux’ might differ in significant ways that could bear on the problem at hand.

  4. I understand perfectly the hurdles dealing with WinXP. I wonder though if finding a single pair of OS Scriveners that play well together and leave the accented characters alone, would be the best solution here. That is: work on Scrivener projects on WinXP and the Mac and leave Linux out of it. Use a special thumbdrive for the transfers - one that never gets anything from the internet installed on it; one that is used exclusively for Scrivener. Or work only on Mac and Linux and leave WinXP out of it (though from your tests it sounds like you are having problems with Linux and that the projects translate fine between Mac and Windows).

  5. This is the ugliest of all solutions, but you could eschew the upper-ASCII characters altogether and employ some sort of hand-coded characters that is easy to read. Then use the replacements in Compile to put the utf-8 accented characters etc back in. A real pain I am sure, but in the old days before utf-8 this was done all the time. Alas that Apple and Microsoft and IBM did not agree on a single extended system and open it up to everybody, back in the 1980s!

  6. Another way to work would be to designate a single system as your Scrivener writing machine and not translate to other versions on other OS’s. Or leave a single project on one machine until it is ready to compile and put into a word processor or submit (that is, Project A would only be on your Mac, and Project B only on the WinXP machine, and Project C only on the Linux machine).

Hope you can get a good working procedure. This is the sort of thing that makes you want to swallow lye, and only takes time away from writing.

  • asotir

I have nothing useful to add to the discussion…

This. Been on deb since … Hmm… slack in 94, opened shop in 95, move to deb must have been 97.

while I moved to mac to get away from maintenance, debian has been very very stable compared to red hat for production use. If you look at the very popular conanical based OS’ you’ll soon see they are nothing more than debian with proprietary add ins. the version of conanical without proprietary add ins are straight up debian in a prepacked form (get kubunto but selecting KDS desktop on debian install). While i tell folks to go conanical it is really only to provide them a support service should they choose to sign up for it. Otherwise go straight debian.