| Commit message (Collapse) | Author | Age | Files | Lines |
| |
|
| |
|
|
|
|
| |
We don't need that anymore since 10.5 support is gone. Sequel Pro now explicitly links Quartz and doesn't try to guess QuickLook support at runtime.
|
| |
|
|
|
|
|
|
| |
certain multi screen configurations
This did affect users with a vertical multi monitor setup using the classical (single Space) multi monitor mode.
|
|
|
|
| |
This is a workaround for a potential data loss/corruption bug until we really fix this (see #1876)
|
|
|
|
|
|
| |
See Preferences > Network.
Note: If you play around with that too much and libmysql can no longer connect, don't hope for a meaningful error. It will rather be something like "protocol version mismatch".
|
| |
|
|
|
|
| |
Hey, I was against this, but if you want it... - just don't blame me for what might happen :shipit:
|
| |
|
| |
|
| |
|
|
|
|
|
|
| |
database (#2082)
(While we're at it, also removed some duplicate CREATE DATABASE code)
|
|
|
|
| |
xibLocalizationPostprocessor robustness
|
|
|
|
| |
Replaced some (range.location + range.length) with NSMaxRange(range)
|
|
|
|
| |
* Can now also reverse the key replacement: Pass in an original IB strings file and a translated strings file in the DMLocalizedNib format and it will generate a translated IB strings file
|
| |
|
|
|
|
|
| |
* This works similar to a regex matching "abc" as /a.*b.*c/ (ie. all characters of $needle need to be contained in $haystack in the correct order but not neccesarily consecutive). Additionaly some unicode equivalencies are handled.
* Changed a tiny helper function from ObjC to plain C
|
|
|
|
| |
The red (first) connection color would not be duplicated to another tab, when choosing "Open Table in New X" from the context menu (issue reported by Kyle via e-mail).
|
|
|
|
| |
when they don't have a label assigned
|
| |
|
| |
|
| |
|
| |
|
| |
|
|
|
|
| |
(This also affects PSMSequelProTabStyle, but see the comment there why it is left unchanged)
|
|
|
|
|
|
| |
When the last selected favorite was "Quick Connect" with SSH, on the next launch SSH would be re-selected, but the fields other than SSH Host would be cut off.
This was caused by a race condition between our window setup code and Cocoas window restoration code. Solved by disabling window restoration for now (didn't really work for us anyway).
|
| |
|
|
|
|
|
| |
* To reduce confusion (and create some, for those people not yet aware of the difference) I have renamed the utf8s to "UTF-8 Unicode BMP" resp. "UTF-8 Full Unicode" according to their supported character range.
* Also *both* utf8 variants will now be at the top of the list when creating new tables/databases
|
|
|
|
|
|
|
|
|
|
|
|
| |
flex does not(*) support UTF-8, therefore
alpha [a-z_\.À-゚]
has always been interpreted by flex as
alpha [a-z_\.\xC3\x80-\xEF\xBE\x9F]
I assume this is not what was indetend and the only reason it worked, is because C3 (195),BE (190) and 9F (159) are already covered by 80-EF (128-239). Incidentally this range would also cover the whole Unicode BMP in UTF8.
This change should make it more obvious.
(*) There were some patches in 2012 and 2014 but they don't seem to have been merged.
|
|
|
|
| |
cause visible changes)
|
| |
|
|
|
|
| |
Voodoo magic provided by @Bibiko ;)
|
| |
|
|\
| |
| | |
Fix an issue with ssh -L forwarding (OpenSSL 6.7 onwards only supports ":" as separator to "-L")
|
| | |
|
|/
|
|
| |
address #2042
|
|
|
|
| |
messaging a Zombie and crashing
|
| |
|
| |
|
|
|
|
| |
fixes #2049
|
| |
|
|
|
|
| |
Depening on your diff tool (Kaleidoscope and Tower are pretty bad, AppCode works well) this change might look big, but all I did was remove two levels of else { } by using return; instead.
|
| |
|
| |
|
| |
|
| |
|
|
|
|
|
| |
Take this commit as a proposal to formalize our existing "[x release], x = nil;" convention by introducing a macro for it.
Feel free to revert this commit if you see issues with the approch or implementation.
|
|
|
|
| |
When cancelling a connection, because SSH is asking for a password, the "SSH connecting…" message in the upper right window corner would not be reset.
|
|
|
|
| |
(Also deduplicated some code)
|