Folder Sync: First impressions

[Tim]

Really? That strikes me as unexpected behavior. Each tab holds a separate session, and they may have nothing to do with each other.

[Michael Bulgrien]

I agree with Ron and Tim. I definitely would not want sides to swap on all tabs!

[ron]

I'm not sure I have that much need for Folder Sync, but maybe I'm not using it right or expecting the wrong things from it.

I was looking for a quicker way to update my local copy of source code. Basically, I most often want to update any files of mine that are older than in the master pool. So I bring up the folder compare, choose Diffs No Orphans, Select All Files, Update Left. I usually deny overwrites and handle any left overs manually. I wish I didn't have to Select All Files. Seems like Update/Mirror would imply selection of all files from one side. Whereas the Copy button would certainly require the user to 'select' some files to process.

I was thinking that Folder Sync might reduce the number of steps. Fire it up from the command line, choose Update Left, then click the BIG button. But, the view setting of Diffs No Orphans is not selectable. Nor is this setting transferred when I choose Sync Base Folders In New View. So the Folder Sync basically means I get everything from the source tree that's not on my machine. Each of my projects usually has only a handful of the 200+ folders present in the source tree.

[Chris]

Hi Ron,

You can avoid syncing orphan files in the Folder Sync.

After loading the two folders in the folder sync session, select "Update Right" as the sync type. Then click the referee icon (Session|Session Settings). Go to the Sync tab and change "if file is only on left" to "leave alone".

This will prevent orphan files from being synced.

[ron]

FYI, it looks like you need to click Accept after changing the sync type before clicking the referee. Otherwise it doesn't seem to work as expected. Bug

In any event, that's the setting I needed. Looks like I can update my defaults but can have only 'one' Custom. So, in my case, the custom will be a customized Update Left. Easier to remember because I'm left handed and the server is physically to my right

[dr_barnowl]

Swap sides :

My position was that sometimes, when using the shell extension to open a folder session, you compare a file, realise that it's the wrong way round, and flip the sides in your file session ; if the flip also propagated into the parent folder session, and into any file sessions that were children of that session, that would seem more consistent to me than having a mixture.

I agree that it would be unexpected for it to flip sides in _unrelated_ sessions, but where the sessions share a common ancestry, I would find it more consistent.

This was partly provoked by the new tabbed UI - the sessions feel more intimately associated with their children (I go with one window per folder session + file sessions started from it). There is a parent/child relationship between sessions in BC2 ; my proposed behaviour does not occur there either, but I think it feels more tolerable because the sessions open in a separate window.

[Michael Bulgrien]

if the flip also propagated into the parent folder session, and into any file sessions that were children of that session, that would seem more consistent to me

I would not be opposed a swap sides that affects such "related" sessions. If implemented, swapping sides in a child file session should also swap the parent folder session and all sibling file sessions. However, it might be tricky (and a lot of work) to keep track of session relationships.

For example: if I launch a child file session...then manually edit the filenames in the file drowdowns to compare another file pair in the same parent folders, will BC3 still recognize it as a child session? Likewise, the manual edit could point one or both sides to a completely different filepath. This would no longer be a child session. What if the user then uses the back button to return to the prior file session. It is once again a child session of the parent folder session. What if the user launches two different file sessions from the Explorer context menu without launching a parent folder session? Should a swap sides work in both file compares since they are siblings (even though a parent folder session does not exist)? The risk for inconsistancy and user confusion is quite high. While an idea for the BC wish list, I would think it fairly low priority since it may take quite a bit of work to make if feel natural.

[Zoë]

However, it might be tricky (and a lot of work) to keep track of session relationships.

We already keep track of exactly that relationship so things like "Next Difference" work.

if I launch a child file session...then manually edit the filenames in the file drowdowns to compare another file pair in the same parent folders, will BC3 still recognize it as a child session?

No. Anything that changes the path edits breaks the child/parent link, as does saving a child session as an explicit session. I think Swap Sides actually breaks the link right now too...

What if the user then uses the back button to return to the prior file session.

File sessions don't have a back button.

What if the user launches two different file sessions from the Explorer context menu without launching a parent folder session? Should a swap sides work in both file compares since they are siblings?

No. The parent/child link only exists for file comparisons launched from a specific folder comparison.

[ron]

Swap sides :

My position was that sometimes, when using the shell extension to open a folder session, you compare a file, realise that it's the wrong way round, and flip the sides in your file session ; if the flip also propagated into the parent folder session, and into any file sessions that were children of that session, that would seem more consistent to me than having a mixture.

I understand the point now, but in this case I'd typically recognize the need for a swap before I had very many child sessions opened. Usually it's 'enter', ooops - wrong way, 'escape', 'swap', 'enter', now that's better.

[Aaron]

So, it is less a matter of difficulty, and more a matter of expected design.

Should Swap sides affect the parent as well?

More discussion on the topic is encouraged. Any other users who would expect for it to be one way or the other?

[Michael Bulgrien]

Should Swap sides affect the parent as well?

If implemented, I would think it should work both ways:

Swapping sides in the parent folder session should swap sides in all child file sessions.
Swapping sides in the child session should swap sides in the parent and all sibling sessions.
(the parent and all of its children)

No. Anything that changes the path edits breaks the child/parent link, as does saving a child session as an explicit session.

My point exactly. A user may expect it to remain a child session even though BC has broken the relationship.

So, it is less a matter of difficulty, and more a matter of expected design.

If you swap a parent folder session, should child folder sessions be swapped as well as child file sessions? Probably.

How many levels deep should the swaps sides be cascaded through parent/child relationships?

I would say that the difficulty would come out of the expected user experience, not the expected design. In other words, if a user expects the relationship to remain when BC breaks it, then logic may need to be introduced to determine whether or not to treat multiple tabs as if they maintained a relationship when, in fact, they have not.

No. The parent/child link only exists for file comparisons launched from a specific folder comparison.

What about folder sessions launched from file sessions using the "Compare Folder Session in New View" option? Is a parent/child relationship appropriately established here as well? I am guessing not, though a user may expect it to be.