Age | Commit message (Collapse) | Author |
|
Change-Id: Ia90a4c371e66f4a5b9a37b4c4ac7d3118ad518a5
|
|
Change-Id: I9965f58b8f06f1cec2c419dcf16d8aebf9cd97b8
|
|
I.e. make it possible for the opengl backend to decide to use a different
scaling algorithm.
Change-Id: I36e76de071797129c7636b3048c93bc6e5d93aca
|
|
Change-Id: I491a6410f70b063fd74834d5837be34e4b9552d7
|
|
Change-Id: Iffe850d99f2dcd9e955a4ba1190e010be905434b
|
|
Change-Id: Ia65754272ff5c1661241dd503cef23c7a89b660e
|
|
Change-Id: I424ac73f23bed5b3782c148d9242ca8fc29d200d
|
|
Change-Id: I3a67da89cedbb6a58b2556abf4553857125af5a6
|
|
I.e. between one PreDraw()/PostDraw() pair.
Change-Id: I358d603ff33fa7416a4033bf074fe390b1112fcc
|
|
Change-Id: I2a2f18d76b0deb5f6cfd68b36699d940703372b3
|
|
One clear bug in the code, in my opinion, was that
PDFSigningPKCS7PasswordCallback() returned its argument as such. However, a
PK11PasswordFunc should return "a pointer to the password. This memory must
have been allocated with PR_Malloc or PL_strdup", says
https://developer.mozilla.org/en-US/docs/Mozilla/Projects/NSS/SSL_functions/pkfnc.html
.
I could not test this fix fully before my hardware token decided to block
itself, thanks to too many wrong PIN attempts. Possibly it would work to even
just pass NULL for the password callback function and its argument to
NSS_CMSEncoder_Start(). After all, at least with the hardware token and
associated software that I tested with, the software itself pops up a dialog
asking for the PIN (password).
Change-Id: I85a8b2833cfdd1a1d7b7779016fefb71dd53ab80
|
|
Change-Id: Ie230115d71bffeb08fc13380648e8216c43f3a5d
|
|
Change-Id: I0ff1e44f17976fd3a3060d8ee7fe15d77c918b79
|
|
Change-Id: I6b1b790bfcc94823f703679689bbd76312ee3e5b
|
|
Change-Id: I44648fb951d03090759d8afaa20cab7d945f684d
|
|
Change-Id: Ie1431f24031265e6fe54c6f0523261da88d71b00
|
|
Change-Id: Iaa0c47bcd53eaaf2577a5401762afaabe44e7514
|
|
otherwise vcl's clipping doesn't work quite right when the render text
with vcl apis fallback is used.
Manually forced in my case, but it should happen in practice with vertical
text, so if there is a bug about vertical text not appearing in slideshows then
this is part of the fix for that.
Windows and Mac remain unchanged as initialized with 1, 1. If the same problem
affects those platforms then they'll need to be adjusted to remember their
height/widths from the ctor and those values plugged in here instead
Change-Id: I2f82f0db0cf446d7db21f0a7ee4f8c15c7ebdb42
|
|
Change-Id: Ifce85b1a7fa8e8b6fc57802a5fb559d2db50ba2a
|
|
As agreed with Adolfo (and others) at the Design hangout on 2014-12-10.
Change-Id: Ibe0e8075566523e746866b9d2c15c348c345d2df
|
|
Change-Id: I0b6fd63f796598e56b2d8abe67df939240019217
|
|
Change-Id: I240d2b44e77d28af3cd5952b6d666a1709c4c54a
|
|
Conflicts:
include/vcl/opengl/OpenGLContext.hxx
Change-Id: Id9c7932976ce9d9282776c20d93d9cca4d290056
|
|
Change-Id: I66496fae902db9df5b8301b00bb779f42adaa7a7
|
|
Conflicts:
vcl/source/opengl/OpenGLContext.cxx
Change-Id: Ie2b93de8efe5ea56b0420adf23639c0153103385
|
|
Conflicts:
vcl/unx/generic/gdi/openglx11cairotextrender.cxx
Change-Id: I469b34c9f1047a274550229391d3dfb578291df6
|
|
Change-Id: I516cdbc466f3d6427e36fea8c5cdbe718ce7d0ea
|
|
Change-Id: Id446d20599f072f657c6106d6c6457fce08830e6
|
|
Change-Id: Iec9f8c7d8f7cded0aef9e017373e44387dc0b05c
|
|
found by PMD
Change-Id: Id6737916b68ccbdbdeec5d314747a38410923ac6
Reviewed-on: https://gerrit.libreoffice.org/13409
Tested-by: LibreOffice gerrit bot <gerrit@libreoffice.org>
Reviewed-by: Noel Grandin <noelgrandin@gmail.com>
|
|
The signature should be in DER-encoded PKCS#7 format and what CryptSignHash()
produces is nothing like that. Luckily CryptSignMessage() is actually almost
easier to use and is capable of doing what we need. This also means that we
won't need any HCRYPTPROV or HCRYPTHASH after all so all the code related to
that can be removed. CryptSignMessage() handles both calculating the hash and
signing it.
One less than ideal issue with CryptSignMessage() is that it needs all the
data to be hashed and signed at the same time, so we need to keep both buffers
around for signing.
It also turns out that we don't need to look up the certificate anew from the
user's certificate store after all.
Now Adobe Reader doesn't complain any longer about the signature's format and
contents.
Change-Id: I25cfb93b516ffa723c6228d068d9ffa8e7cc4790
|
|
There was one details that I had missed in my initial coding:
CryptAcquireContext() doesn't give you a HCRYPTPROV key container that
would contain the private key of a public key certificate. For that
you need to use CryptAcquireCertificatePrivateKey(). When the hash has
been created using such a HCRYPTPROV, the CryptSignHash() call
succeeds.
The certificate in DER encoding that is passed in from the caller,
obtained in the certificate chooser (in xmlsecurity), is possibly not
good enough to be used for the other things. So look the same (?)
certificate up in the user's key store instead. At least more
properties are present in the certificate when looked up like that.
Add more SAL_INFO logging, with cleartext dumping of certificate
context property names and list of algorithms supported by the CSP.
Unfortunately, even if all the WinCrypt API calls now succeed, the
signatures we produce still are not good enough for Adobe Reader... A
lot of information must be missing, they are quite short, just 256
bytes.
Change-Id: Ifa4dd37b6d40932fcdcbb07e00c9eb52d54a5477
|
|
Change-Id: I30bf474fc44b122c8b7218e187c822a8784783fd
|
|
Change-Id: I661949a8635a2ed1a1efda77ced48a0fa9f6722e
|
|
Change-Id: I3ae51b8c57e70593153805e5c7c139573c543b5a
|
|
It is already checked on all call chains:
* CreateSalInstance (vcl/unx/generic/plugadapt/salplug.cxx) checks
IsHeadlessModeRequested before calling autodetect_plugin ->
get_desktop_environment -> get_desktop_environment
* Application::GetDesktopEnvironment (vcl/source/app/svapp.cxx) checks
IsHeadlessModeEnabled before potentially calling SalGetDesktopEnvironment
(vcl/unx/generic/plugadapt/salplug.cxx) -> get_desktop_environment
-> get_desktop_environment
* The two calls to SalGetDesktopEnvironment in vcl/unx/gtk/ are clearly only done
for GTK plugins, not headless
Change-Id: I44b43cecd63effc5b2e3ad2a449943fee183462a
|
|
Change-Id: Ib459799f4a3224f8c9683ac4b6cf37982d2077a3
Reviewed-on: https://gerrit.libreoffice.org/13406
Reviewed-by: Noel Grandin <noelgrandin@gmail.com>
Tested-by: Noel Grandin <noelgrandin@gmail.com>
|
|
Change-Id: I4d15a0e606f468346bc5d2f43af2e14bcc758e13
Reviewed-on: https://gerrit.libreoffice.org/13405
Reviewed-by: Noel Grandin <noelgrandin@gmail.com>
Tested-by: Noel Grandin <noelgrandin@gmail.com>
|
|
Change-Id: I9bfd611ef5451b936f2b7b4dda6dd8ede23832a4
|
|
Change-Id: I70a769bac1fa04ff5606b729772c1d28b144b64e
|
|
Change-Id: Ica56b6b07f8fc521882d15c1bf95cc6c162b4fd1
|
|
Change-Id: Iae2c642ad992c822720bcf1e34827bfaa35876b6
|
|
Change-Id: I1d70d4d2fe8573d0f120f2a1ece83f504d5efde5
|
|
Change-Id: Iaceef008e7df91f3a9ad54fdaec25f2170100ee0
|
|
Change-Id: I05ccb8bd73070462edcc911956859aa7967b5901
|
|
Change-Id: I57f22b6d9bfbef4a50d162076b01e8e64edb4718
|
|
An idle handler will ultimately be a zero time timeout with
prioritisation layered on top of that.
Change-Id: I3f0802d5001172fc7b8409274bc5a3632e5dad34
|
|
Change-Id: I0d2119a297c9a6d9b99676a561ddf4dd654ecaea
|
|
Change-Id: Iac4b100e3eca456fdd19d67ff311a02c40dc060a
|
|
Change-Id: I5846940cc45781f4a0264047107dbb568604d543
|