Project History panic

Question

Today, I wanted to create some text for a colleague to try out in SAB (scripture app builder). This is what I did in PT8:

• create a copy of the book (LUK) in the windows explorer
• open PT8 and make a manual “project backup to a file”
• create an entry in the project history (named “Friend”)
• select 40 Biblical Key Terms which have glossary entries assigned
• use the tool “Mark Occurrences of Selected Terms as Glossary Item” .
• this generated a lot of tags in the text
• copied the LUKproject.sfm file for the colleague
• found the “Friend” in the project history
• asked PT8 to revert everything to the “Friend” state
• found that none of the many \w and \w* have been removed
• did ripping of garment, gnashing of teeth, broke sweat, got grey hair, remembered this forum…

I know from previous experience that glossary markup is painful to work with while actively translating and I have notes to self to only ever mark up glossary entries as one of the last states before publishing. This is why I made a project history entry and made a manual backup.

This project relies heavily on send/receive as it involves an entire team and four computers.

Questions:

• Where did I do wrong? Does the project-history-revert not cover “everything”? Is markup like \w not un-doable?

• Assuming it was somehow my fault, how can I manually re-introduce a manual backup file like 43LUKproject.SFM into the project, without “confusing” the inner management (like send/receive) of PT8 please?

• Or can I use the “backup project to file” to savely re-gain the status of this morning without risking any side effects with our send/receive setup. (I happen to be the (only) project administrator for this one.)

Please do not tell me “you do not want this”. Those markers have to go. They were added for an external app-making-experiment. This language project has very complex glossary entries, and since some of the key terms have several renderings (and not each of those applies to the assigned glossary entry) we have a definite need to undo.

For those who want even more, an example: The Greek κατακλίνω is sometimes rendered as “recline at table” and this needs some explanation for the local readers. But this explanation would not be helpful at all, at all the locations where the Greek term was just translated as “sit” or even “eat”.

Answer 1

Tim, the process you describe should have worked (reverting). I would suggest using the history to look for the point where the \w…\w* do not appear in the text.

As a side note - it is very dangerous to restore into an existing project using Backup/Restore - this will likely introduce conflicts to the text. It is better to use the project history. You can, however, use Backup/Restore to create a new copy of the project.
My experience is that sometimes I have to look carefully at the text to see which point I want to restore to. Creating a point should normally work - it did in my test of your process.

A second side note - if you want to create some side project for someone else, it is better to create a copy of the entire project and work with that project.

Blessings
anon848905

Comments

anon848905:

the process you describe should have worked (reverting)

I have done some digging around and tried again and can report that the panic is gone, the project data looks now restored to the “point in history which I had created specifically for my markup experiment”. Thank you all who gave helpful input.

What has happened? The best I can tell is that my first attempt at restoring has happened but did not work. The online help is saying “A point in the project history will automatically be marked for you just before this procedure”. I can see this confirmation from last week Friday, but it only shows 1 change.

I have just now managed to do a working restore to my history point of Friday morning and its confirmation is showing 166 changes, which is more fitting (no content-work has been entered since the panic).

Funny detail: I can see that exactly ten minutes ago I reverted, but I have not found where I can see to which history-point this revert was pointed. Do I need to learn another mouse-hover trick, late where the mouse-hover over the date shows the time?

From now on I will heed @anon848905 's advice to create a project-copy for any such transmogrifications for publishing purposes or for experimenting.

Still, I want to better understand the “normal project history” procedures. The window, where users create such history points, is giving this advice: “We recommend you mark a point in the project history both before and after making any major change in a project.” (emphasis is mine)

The pih “before” is obvious. And last Friday the pih “after” I had not done, because the plan had been to revert right back when I was done copying the raw text for my colleague.

Is the recommendation on a pih for “after” a major change technically important somehow?

So maybe I need to know more about the “inner concepts”. Should I mentally consider a “point in history” as no-longer-changing-batch-of-data-with-a-time-stamp or rather as a-fresh-page-on-the-project-log-with-a-time-stamp-at-its-beginning? Or can somebody give a better mental illustration?

In other words, can I only properly revert to a point in history after I have created another younger point in history? Or could I create a pih, make one change in the text and revert right back?

---

Tim:

“We recommend you mark a point in the project history both before and after making any major change in a project.” (emphasis is mine)

The pih “before” is obvious. And last Friday the pih “after” I had not done, because the plan had been to revert right back when I was done copying the raw text for my colleague.

Is the recommendation on a pih for “after” a major change technically important somehow?

I’m sure you’re right that that the “before” point is the more important of the two: it allows you to go back to where you were if you do a disastrous Find-and-Repace, for example.

But if the next thing you do is also a disaster (accidentally delete a large amount of text, or whatever), it’s nice to have a point after your Find-and-Replace so that you can Compare Texts or Revert Books and only be dealing with the latest changes, not the latest changes plus the Find-and-Replace that happened just before.

Even if there’s no “disaster”, sometimes you need to just Compare Texts to view the changes somebody made – their regular edits to the text – and you want to be able to do that without having a whole bunch of other changes form a Find-and-Replace intermingled with the edits.

So maybe I need to know more about the “inner concepts”. Should I mentally consider a “point in history” as no-longer-changing-batch-of-data-with-a-time-stamp or rather as a-fresh-page-on-the-project-log-with-a-time-stamp-at-its-beginning? Or can somebody give a better mental illustration?

It’s a snapshot of the text at a given point in time.

But note that if user A does some edits today but does not S/R, and user B does some edits tomorrow and S/Rs immediately, then you S/R, you will see the later chages but not the earlier ones. This is because, until user A does an S/R, the changes are not on the server, and not visible by others.

When user A eventually does do an S/R … this is where it can get complicated … user A’s Paratext will have done a commit of the text at least once every 24 hours unless there were no edits (see What is a “commit”? (and what is Mercurial?) ). Each commit will become a point in the Project History, presumably creating some points with timestamps earlier than some points already recorded in the history before user A’s S/R.

In other words, can I only properly revert to a point in history after I have created another younger point in history? Or could I create a pih, make one change in the text and revert right back?

No, I don’t believe there needs to be a younger point before you can Revert: Paratext also recognises that there is a “Current Version”, i.e. the state of the text right now on a given computer. You can see this point in a Compare Texts window, but not in the Project History (since the Current Version is always changing, and it’s impractical to record every small change. That’s why PT does a commit (mentioned above) to the Project History once a day.

---

Sorry, that’s quite a long answer, but I hope it’s helpful. One of the developers may correct me if I’ve got anything wrong.

---

wdavidhj:

One of the developers may correct me if I’ve got anything wrong.

Looks good to me.

See also this post I made a while back attempting to explain why Project History (specifically Verse History) is complicated.

Answer 2

Hi Tim,
I will let the real gurus answer your question about why marking a
point in the project history didn’t do what you were expecting it to
do, but this is also something that we have had to deal with in the
past. If you are just wanting to carefully rip out all the \w
keeps-this|but-delete-this\w* markup, then this
link from our workshop notes document might help you to deal
with them using a Regular Expression.

Mark

Comments

Thank you @Mark+P for your idea to directly remove the unwanted markup via regexes. This had occurred to me a few days ago in the shower. I love regexes and will compare your link to my own syntax.

We were travelling and I am only now back to my own desk. I will do a bit more research and will try to understand the project history concepts and what went wrong. But whatever I find, I am glad that you confirmed that we have a working solution to get rid of this batch of test-markup.

I will also look into the process of setting up copies of projects for publishing-extracts-while-the-main-translation-branch-keeps-moving-forward-unendangered.