Age | Commit message (Collapse) | Author |
|
xtext keeps a static buffer and uses it for various things
and asserts that every text entry is < 4096. It does
this check on gtk_xtext_append*() except it does the check only on
the right half of text when indent is enabled.
This overflow caused corruption in the xtext struct
changing the url check functions making hovering
with the mouse do 'undefined' things.
In the long term this should be removed for a dynamically
allocated buffer so no arbitrary size limit exists and
text gets cut off.
Fixes #1465
Fixes #1186
Fixes #1206
|
|
|
|
|
|
|
|
the rest of HC.
OS name, CPU, GPU, HDD info and build arch are now all available within core HC. OS name and build arch are used in the About dialog.
|
|
|
|
|
|
Also cache the result...
Fixes #1500
|
|
https://bugzilla.redhat.com/show_bug.cgi?id=1262763
|
|
|
|
I'm removing Seion from the serverlist because it's now mostly a defunct network. One of the servers is completely broken in regards to ipv6 connectivity and SSL, the webchat server hasn't worked for months, and I shut down my server (fox.seion.us) a few days ago.
|
|
This fixes /mode using the name of invalid contexts
and also adds some documentation to what is happening.
In the end though it still must guess between modes
and nicks if ran in a valid context.
Fixes #1470
|
|
Fix selection when part of the selected text is above/below a border.
Also, fix problems uncovered by static analysis.
Fixes #1261.
|
|
Closes #1438
|
|
gtk_entry_get_text_length() returns len in utf8
chars not in bytes
|
|
|
|
Alt+a is used for mark away in the menus and also
highlight all in the search box, change the latter to
alt+h.
Fixes #1008
https://bugzilla.redhat.com/show_bug.cgi?id=1173082
|
|
This fixes xtext re-rendering multiple times when the
indent width is not sufficient
Fixes #1364
Closes #1367
|
|
|
|
gtk_entry_get_text_length returns the number of utf-8 characters in the GtkEntry, not the number of bytes in its text.
|
|
Fixes #1360
|
|
No network can agree on what mode r actually
means so we can't have nice things.
Closes #1287
|
|
This no longer restricts their width and uses
monospace fonts so they are still the same
width.
Closes #1356
Fixes #1354
|
|
Closes #1265
|
|
Closes #1343
|
|
|
|
gsize and size_t are not the same thing
|
|
It still has to be in hex...
|
|
|
|
|
|
Fixes #1148
|
|
No reason to roll our own.. in multiple places
|
|
It handles the correct encoding on win32
|
|
Thankfully somebody already made a define for this, they
just never used it...
|
|
|
|
not working.
Fixes #602
|
|
Proper fix for b66edd3
This issue likely applies elsewhere in HexChat
|
|
|
|
|
|
URLs to open them.
This reverts commit 8ca96d7ea3ec0281802eb08174ca2fdcbcc8263f.
|
|
Distros like Debian depend on repeated builds
having the same results which this breaks.
Since this provides no real value just
remove it.
|
|
|
|
Fixes #943
Closes #1293
|
|
openssl seeds itself with CryptGenRandom on Windows in the same manner as /dev/(u)random on other OSes.
|
|
compression support.
The function is still defined in openssl/ssl.h but not in the actual library, so it cannot be linked to.
|
|
Fixes #1282
|
|
This setting doesn't work on most platforms
including some notification daemons on linux.
Where it does work I think it is bad behavior
to default overriding their defaults anyway.
It's most valuable use was likely win32
tray balloons which no longer exist.
|
|
- Output directly to rel\ instead of to bin\ and then copying files over.
- Deduped Win32 vs x64 configs
- Moved some common properties to hexchat.props
- All build intermediates (except htm's intermediates) are no longer emitted in the source directory
|
|
process, not from a DLL.
They're process-level functions and aren't meant to be called from DLLs. CoInitSecurity in particular fails with RPC_E_TOO_LATE even if no other call to CoCreateInstance has been made yet.
Fixes sysinfo's WMI calls on Windows 8.1 and above.
|
|
This sadly can't work everywhere since they all
have entirely different ways of saving the data..
Related to #1288
|