A minor caveat is that if there are encoding or newline (folder specific) settings in effect for the chosen path, these trump what’s selected in the save dialog.
If we wish to solve this, the best would be to update the options (shown in the save panel) based on selected folder/filename (and the settings in effect for that).
This closes issue #163.
We moved it to TMPDIR so that multiple instances of TextMate could run (for different users) but have since gotten a lot of reports about problems with this location. The short version is that ‘ls’ shows the socket exists (in TMPDIR), but stat’ing it or similar will fail.
To still allow different users to use mate side-by-side we add the user ID to the socket name.
Hopefully this closes issue #183.
The reason for this is that we might be able to overwrite mate even if we do not have write access to the parent directory. This will avoid us having to prompt the user for an admin password (when updating mate).
It should be mentioned though that incase mate is not a regular file, we do need to remove it first, because if e.g. it is a symbolic link, our install would write to where the link points to.
Theme was no longer initialized by OakTextView, instead we relied on OakDocumentView to provide us with a theme, but if the user had no themes installed, it would never do so, leading to a crash when OTV uses a theme member function to obtain colors.
This closes issue #222.
This was (temporarily) disallowed by previous commit. We do however need to allow it in the last placeholder since that is used for chaining snippets. See issue #182.
If caret is at the first position of a place holder then we no longer look for tab expansions, should the user press tab.
This address the issue where a placeholder has optional content and the word to the left of this optional content is a tab trigger. Previously deleting the placeholder text and pressing tab would insert a new snippet, which would generally not be expected.
This closes issue #226.
It can happen that we register the same value transformer twice, e.g. if we register it in an initialize method (since these are not guaranteed to only be called once).
This isn’t a problem per se, but there is no reason to create the new instance.
Previously we used ‘dealloc’ to release the array, which is wrong.
Since instances of OakStringListTransformer is normally never released this bug wasn’t affecting anything. But since we have a dealloc, it should be correct.
This will expand/collapse the find results.
The benefit here is just having a key to quickly collapse all the results so that we get a list of files (with matches).
This is to avoid running out of memory, ideally we would search in chunks, but given that Oniguruma has no direct support for this, such fix is more involved, and it’s likely rare people want to perform regular expression searches on files larger than 5 MB, so I feel this is sufficient to close issue #38.
Ranges can be unanchored which is when they are made (mainly) by a unit selection, for example “select to enclosing braces”. What it means is that if we later use shift + movement actions (to “move” the selection end point) we will always extend it (as we don’t know which of the two end points should be considered the anchor).
Though the unanchored flag has a double role, for snippet placeholders we also set it, when we select the placeholder field, and when user press tab, we ignore tab expansions if there is an unanchored selection. So this disables recursive snippet expansion if the user immediately press tab when reaching a snippet placeholder.
Turns out that when we reach the last field of a snippet, some users want a new snippet expansion.
This fix is not ideal, but by far the easiest, and it achieves the goal.
This closes issue #182.
It now takes key path and destination (github repostiory) as arguments. The generated build.ninja assumes the user local build.ninja sets these variables.
This means pressing e.g. ⌘/ (using the slash on the numeric keypad) will still work to toggle comments.
It also means that we can no longer bind specifically to the numeric keypad keys, but I don’t think anyone actually does that (given that few have numeric keypads).
The ideal solution would be to first do a literal check and then, if there was no match and the key event had the numeric keypad flag set, do a test with this flag removed, but that would change the simple string compare we presently use plus the easy caching and binary search of finding the item that matches a key event, i.e. it would add complexity to the code with no known argument in favor of this flexibility.
This was also done prior to commit 806cb44. I just assumed that now that we are back using the system’s named image stuff, we would benefit from their cache, but clearly we are not (especially file browser seemed sluggish because of (re)loading the file type images).
Previously it would place either the column or discontinuous selection on the find/replace clipboard, which wasn’t useful.
Now it only places the “first” range (of a discontinuous) selection on the clipboard, or first row of a column selection.
When we need to work on a paragraph (reflow, unwrap, select, etc.) we now extend upward until we see a blank line, and likewise move downward until there is a blank line.
This is related to issue #154.
