Having a separate mutex for the three or four instructions needed to
fetch and increment nextID means the overhead exceeds the cost of this
operation. nextID is now handled inside the critical section for
awaiting instead, while the more expensive channel creation has been
moved outside it.
This is mostly a simplification, though it may have minor performance
benefits in some situations. The single-threaded sender benchmark shows
no significant difference:
name old speed new speed delta
RequestsRawTCP-8 55.3MB/s ± 7% 56.6MB/s ± 6% ~ (p=0.190 n=10+10)
RequestsTLSoTCP-8 20.5MB/s ±20% 20.8MB/s ± 8% ~ (p=0.604 n=10+9)
Add a link next to each setting's label to its explanation in the Docs.
This way, it is easy to quickly check what the setting is about without
going to the Docs site separately and searching for it manually.
Signed-off-by: Tomasz Wilczyński <twilczynski@naver.com>
Right now, the Trash Can versioning info text is displayed for both the
Trash Can and Simple versioning. However, the Simple versioning has its
own info text that is also displayed, which results in two very similar
sentences being shown on the screen.
This commit limits the Trash Can info text to be displayed only when the
Trash Can versioning is selected, and moves the Simple versioning info
text up to the same location as the other one.
Signed-off-by: Tomasz Wilczyński <twilczynski@naver.com>
Even though technically possible, CJK languages normally don't use
italic text at all, as not only does it make the characters/letters look
unnatural, but also, in the case of complex characters, unreadable too.
For these reasons, it is usually recommended not to use the italic font
style at all [1][2].
This commit changes the default font-style of the i element for Chinese,
Japanese, and Korean langauge to "normal" instead of "italic". In order
to do so, the HTML lang attribute is also changed following each change
of the GUI language.
[1] https://bobtung.medium.com/best-practice-in-chinese-layout-f933aff1728f
[2] https://devblogs.microsoft.com/oldnewthing/20060914-02/?p=29743
Signed-off-by: Tomasz Wilczyński <twilczynski@naver.com>
Staggered File Versioning used to have its own cleanInterval that
controlled how often file versions were cleaned. Nowadays, there is a
seperate setting called cleanupIntervalS responsible for the cleanup,
which applies to all File Versioning (except External). Thus, remove the
unneeded code and don't set the param up on new folders anymore.
Signed-off-by: Tomasz Wilczyński <twilczynski@naver.com>
* gui: Allow to translate and fix incorrect Versions date filter ranges
Translate the previously English-only ranges used in Versions date
filter. In the process, fix the currently incorrect range calculation.
For instance, let us say it is 08:05. Selecting "today" should set the
range to start at 00:00 and end at 08:05. However, what really happens
is that both start and end are set to 08:05, and as a result "today" is
never shown to the user. The case is the same for "yesterday", which in
contrary to "today" is shown, but its range is fixed at 08:05 on the
previous day.
This commit fixes the above by always calculating "today" starting at
00:00 up to the current moment, and calculating "yesterday" from 00:00
on the previous to 00:00 on the current day.
When it comes to "last x days", the commit fixes the calculation so that
it actually covers the whole day, which is done by moving the start date
to 00:00 on the first day.
Signed-off-by: Tomasz Wilczyński <twilczynski@naver.com>
The ignoreDelete warning code originally comprised of two sentences,
which in total were much longer than the current one. Because of that,
some additional fixes were added to make it present itself better, e.g.
white-space was set to normal to allow for text wrap on narrow screens,
and the help link was moved to a new line to make sure both the anchor
and the question mark always stayed together.
However, the code fixes remained despite the final text being much
shorter than the original. Therefore, remove the unnecessary white-space
formatting, and also move the help link right next to the warning. This
makes it clear that the link is related to this particular warning when
both it and the ignore patterns warning are displayed at the same time.
Ref: https://github.com/syncthing/syncthing/pull/8054#issuecomment-978148117
Signed-off-by: Tomasz Wilczyński <twilczynski@naver.com>
* gui: Translate fancytree messages in Versions modal
Currently, the default fancytree info/error messages are used. This
means that a) they are English-only, and b) they are very generic. With
this commit, the messages are added to the translatable strings, and
they are also more specifically related to file versioning.
On a side note, the "moreData" string has been left out on purpose, as
it is not used in the current code. It can be added later if needed.
Signed-off-by: Tomasz Wilczyński <twilczynski@naver.com>
* gui: Make ID read-only and hide nearby devices when adding pending devices (fixes#8083)
Currently, there is no distinction between adding "new" and "pending"
devices. For this reason, the user is always presented with a list of
"nearby devices" to choose from. This commit adds such distinction to
the code, and in the case of "pending" device, both the device ID is
made read-only and the nearby devices list is hidden.
As a by-product of the function rename and clean-up, this commit also
hides the non-functional "remove" button that is shown when editing
device defaults.
Signed-off-by: Tomasz Wilczyński <twilczynski@naver.com>
The intention was that if no peers are given, we shouldn't start the
listener. We did that anyway, because:
- splitting an empty string on comma returns a slice with one empty
string in it
- parsing the empty string as a device ID returns the empty device ID
so we end up with a valid replication peer which is the empty device ID.
* lib/protocol: Require at least 3.125% savings from compression
The new lz4 library doesn't need its output buffer to be the maximum
size, unlike the old one (which would allocate if it weren't). It can
take a buffer that is of a smaller size and will report if compressed
data can fit inside the buffer (with a small chance of reporting a false
negative). Use that property to our advantage by requiring compressed
data to be at most n-n/32 = .96875*n bytes long for n input bytes.
* lib/protocol: Remove unused receivers
To make DeepSource happy.
* lib/protocol: Micro-optimize lz4Compress
Only write the length if compression was successful. This is a memory
write, so the compiler can't reorder it.
Only check the return value of lz4.CompressBlock. Length-zero inputs
are always expanded by LZ4 compression (the library documents this),
so the check on len(src) isn't needed.
