Document comparison feedback

[Timur Born]

Hello.

I just tried the Document Comparison feature and noticed several things. But first of all: Thanks and congratulations for implementing this! I only just noticed the feature due to a forum post. Great stuff!

- Protected documents cannot be compared?! I wonder which of the following permission details keep comparison from working?


- The original comparison run is single-threaded only, leaving 23 if my 24 logical cores unused.

- The popup about the comparison being in progress does not pop up until I click the XChange Editor window.

- The "Compare Documents" window does not remember the last comparison, but instead tries to compare the tab left of the current one to the current tab. I assume that this is on purpose and I can think of arguments pro and contra both implementations.

- Flipping pages in comparison view has one pane lag behind the other (out of focus side lags behind in focus side), sometimes considerably when large images are present.

- Comparison get easily confused by multiple columns. If a text edit leads to a sentence switching columns then the rest of the page may be listed as changed, even when only a single sentence may have been deleted.

- Some comparison output makes no sense. This is one page where a sentence switches columns, all text afterwards is marked as numbers?



- Somewhat simplistic filtering. My documents have the same header and footnote on every single page, part of that is a publishing date. Unfortunately these do not seem to qualify as what Editor sees as header and footers, so they cannot be filtered out. As a consequence I get differences reported for this same header/footnote on every page of the document with no other way to ignore them.

- Pages should maybe better be aligned to the middle divider line instead of being aligned to the center of their window?!


- No way to switch Old/New documents sides (left/right)?

- "Show all differences" never turns off and I don't know what this option is supposed to do?!

- "Show summary" keeps the "Differences" focus on the last viewed difference. Clicking it again does not reopen the corresponding page, because Editor still things it was showing this page already.

[TrackerSupp-Daniel]

Hello, Timur Born

To keep this short, I'll highlight items I can directly answer here, the rest, both bugs and suggestions, will be forwarded to the Dev team for review.

As a general statement, Document Comparison is much earlier along its development cycle than most features we have ever released to the public. It was released in this state due to such extreme demand that we simply could not hold it off any longer, there is still a VERY large amount of work to do, and it WILL be a long term project. You should expect to see slow improvements over time with this feature, much like you did with EOCR, and 3D before this.

- Protected documents cannot be compared?! I wonder which of the following permission details keep comparison from working?

If a document is protected in any way, comparison will not work on it, as we need to be able to modify, copy, open, view,etc (handle) every aspect of the document for the comparison, yes, at the moment the comparison is barebones, but over time more and more items will be added to be compared, further complicating this.

- The original comparison run is single-threaded only, leaving 23 if my 24 logical cores unused.

This is still an early edition of comparison, improvements will be coming, including use of multiple cores, if it provides any notable improvements in testing.

- Some comparison output makes no sense. This is one page where a sentence switches columns, all text afterwards is marked as numbers?

The comparison output should use the actual text present in the document, if you are seeing numbers where there should be text, you may want to check the documents themselves, in all likelihood, if you select those passages of text, and change the font, the text will turn into numbers. (If that is not the case, please send us an email separately for this issue, with sample documents that we can investigate).

- Somewhat simplistic filtering.

This one needs special attention, At the moment, we only compare base content items, primarily text, and position. Headers are base content within a special container, we have not yet developed the logic to differentiate these items, so the filters are moot for the moment.

The Dev team will be reviewing this topic as well. If they have questions for you, we will ask them here or by email. Thank you very much for the analysis, this helps us to get the feature set smoothed out and ready for more people to use it.

Kind regards,

[Timur Born]

I had another look. The numbers are on the page, but the formatting confusion between both versions makes them appear as differences.

[PHK]

I have not spent enough time on this new functionality to have formed many sensible observations but I am glad you have floated the beta version for us to kick around and it is a feature I would like to see "permanent."

But one threshold comment: I don't like the "old" and "new" designations although I don't have a better idea.

[TrackerSupp-Daniel]

Hello, guys!

Vasyl just got back to me on this and had a minor correction, of the three properties set in your screenshot Timur, the only one that the Comparison actually needs is "extract pages" permission. The other two don't particularly matter, so my statement that "any" security will stop it from working was not entirely true. Sorry for that.

But one threshold comment: I don't like the "old" and "new" designations although I don't have a better idea.

We actually had a brainstorming session trying to find better ideals for these options, and came to the same conclusion, although they may not be "the best" descriptors, they are about the best we could come up with, and they are appropriate for the purpose of comparison, but maybe we will find something better in the future..

Following that, some extra points after Dev communication:

- The popup about the comparison being in progress does not pop up until I click the XChange Editor window.

We cannot reproduce this issue so far on our end, the progress window appears to show up immediately for us. Are you able to reliably reproduce this issue?

- The "Compare Documents" window does not remember the last comparison, but instead...

This is apparently one item where we mimicked Acrobat's handling.
1. When user has opened doc-tabs - we use that docs to 'prefill' the dialog
2. Otherwise: dialog loads the last used docs when shown.

- No way to switch Old/New documents sides (left/right)?

This has been confirmed as intended, and will not be seeing implementation.

- "Show all differences" never turns off and I don't know what this option is supposed to do?!

Currently, this is not a toggle button, and as all differences are shown by default, it would not make much difference to be pressed. We are going to look at making this a toggle instead, to "show/hide" all differences.

- "Show summary" keeps the "Differences" focus on the last viewed difference. Clicking it again does not reopen the corresponding page, because Editor still things it was showing this page already.

This one has been reproduced, and will be fixed in the next build!

Kind regards,

[Timur Born]

- The popup about the comparison being in progress does not pop up until I click the XChange Editor window.

We cannot reproduce this issue so far on our end, the progress window appears to show up immediately for us. Are you able to reliably reproduce this issue?

Turns out that this is a conflict with the third party software Groupy. Once I disable Groupy the popup shows up immediately.

- No way to switch Old/New documents sides (left/right)?

This has been confirmed as intended, and will not be seeing implementation.

Rather unfortunate, because I already unintentionally opened new/old on wrong sides several times and a simple switch would be helpful. Even more so when the original dialog does not remember the last documents being compared. Lots of clicking and waiting for just swapping sides.

One more observation: the comparison process (popup window) can only be cancelled after the progress bar reaches over 40%.

[TrackerSupp-Daniel]

Hello, Timur Born

The problem is that is is not just swapping sides, right now, you only see "moved from X to Y", "removed", "added" or "text replaced" for the base content items, but over time more possible difference types will likely be added. Each of those changes is a unique thing, and it is no simple task to flip the documents from A to B without reassessing the entire document again to offer the correct terminology. And, since changes were already made to add the comparison results pages, there would be further discrepancies if we re-analyzed the new versions of the file instead of the original, which we as we do not lock, we cannot confirm if it has been left unchanged.

In the end, it is simply overly convoluted to "just swap sides", and so it will not be happening. If you need to switch sides, you will need process another comparison of the files, with the "old" file in the left panel.

Kind regards,

[Timur Born]

A button to swap sides could do just that, reprocess the comparison without the user having to enter the source documents again. We do not always compare open tabs and even then it's not always neighboring tabs, so a simple way to save all the clicking for a side swap would be useful.

[Timur Born]

I just wanted to suggest an compromise to handle this without much additional coding on your developers part. Turns out that this is already implemented exactly the way I was about to suggest.

When the "Compare Documents" dialog is opened while the comparison tab is in focus then the dialog *does* remember the last compared documents and then we can simply click the side-switch button in that dialog. A new comparison is then processed and opened in a new tab. So now we have two tabs with both sides switched without having to choose the source documents again.

Still more clicking than a simple button, but better than originally thought.

[TrackerSupp-Daniel]

Hello, Timur Born

Glad to hear that you have found a solution that works for your needs there.

Kind regards,

[PHK]

...

But one threshold comment: I don't like the "old" and "new" designations although I don't have a better idea.

We actually had a brainstorming session trying to find better ideals for these options, and came to the same conclusion, although they may not be "the best" descriptors, they are about the best we could come up with, and they are appropriate for the purpose of comparison, but maybe we will find something better in the future..
...

Temporal juxtaposition is only one of the possible documents' characteristics; perhaps just basic "left" and "right" would serve. But just to complicate design considerations furhter, how about giving the user the option to customize?

[TrackerSupp-Daniel]

Hello, PHK

Funny suggestion, I can say with confidence that it is highly unlikely we will offer a redundant option somewhere in that background that only serves to change the text used. If someone really cares that much about the wording used, we offer our UI translation tools, which can be used to modify your own "custom English" language pack if you so choose.
Also, from a comparison standpoint, "left" and "right" while positionally correct, offers no indication of which the "old version" of the file is. When comparing we need to know which is old and which is new, in order to present the correct "changes" (eg. "text added" vs "text removed"). If we simply use left and right, that would be less clear than it already is.

Kind regards,

[PHK]

..

Funny suggestion

...

ha ha

[TrackerSupp-Daniel]

[PHK]

...When comparing we need to know which is old and which is new, in order to present the correct "changes" (eg. "text added" vs "text removed"). If we simply use left and right, that would be less clear than it already is.
...

One last stab at this: The temporal juxtaposition ("old" and "new") may not be appropriate in all circumstances. For instance, multi-user situations rather than single-user. Suppose you and I are working on an important legal document and I prepare a first draft PDF and send it to you. You add and subtract some verbiage and send it back. OK, no doubt about which is "old" and which is "new." But suppose then that we both start working on drafts but not at the same time. In fact, you finish before I do. Now what relevance is old vs. new? Mine vs. yours might be more relevant. Or Sally's vs. Jim's.

Or if I want to create a sale contract template that is based on a particular house sale but I want to adapt it later to other similar transactions. Time again may not be the crucial differentiating factor; it might be the property or the jurisdication or many other variables. Irrelevant time designations can only lead to confusion.

You can do better than old and new.

[TrackerSupp-Daniel]

Hello, PHK

Easy solution, for familiarity among the majority of new users rapidly fleeing adobe, we do happen to have the same scheme as they do, old on the left, new on the right, I do not have a copy of it, but i believe Bluebeam uses the same terminology for their comparison tool:

For the foreseeable future, we will continue using Old and New, if nothing else, for consistency among users coming from other software.

Kind regards,

[Timur Born]

When you improve filtering then please allow to ignore results. In my sample document I have a simple date change four times on every page out of over 640 pages. Since I currently cannot filter/ignore that it has hard to find any meaningful changes at all.

[TrackerSupp-Daniel]

Hello, Timur Born

Thank you for the suggestion, It would be much more complex than you may expect to filter these items in an intelligent and automatic way, but we may be able to offer a strict manually configured "ignore/hide results which match this exactly". I have created a formal feature request for this, but I should warn you that this is considered a long term request, and likely will not be arriving any time soon.

RT#6149: FR: Comparison tool "hide changes" options

Kind regards,

[Timur Born]

Thank you. Filtering by exact match (with some means to reset) would be very helpful in cases like this example.

[PHK]

When you improve filtering then please allow to ignore results. In my sample document I have a simple date change four times on every page out of over 640 pages. Since I currently cannot filter/ignore that it has hard to find any meaningful changes at all.

This illustrates to me how confusing "old" and "new" designations are in this context. I see "old" associated with more recent (newer?) text changes than the "new." My old head spins.

[TrackerSupp-Daniel]

Hello, PHK

Ideally, you should place the older version of the docuemnt in the "old" slot, and the newer version in the "new" slot, to avoid that confusion before it starts.

Kind regards,

[Timur Born]

Personally I would prefer just left and right, so that I don't have to remember or look up which is which after not using the comparison for some time. But it's not very important.

[TrackerSupp-Daniel]

Hello, Timur Born

The distinction between old and new is important for the presentation of information, not everything is "replaced" many cases there is "removed" or "added" and not knowing which is the "old" file prevents that from being presented properly. As such, OLD and NEW will continue to be used for this dialog for the forseeable future.

Kind regards,

[Timur Born]

Like I wrote before, it's not too important for me. Personally I usually compare left vs. right and have a hard time remembering where I put the "old" one vs. the "new" one. But there are bigger problems to be tackled.

[Paul - Tracker Supp]