Re: obsoleting ONE of the windows build systems

Am 08.08.2016 um 16:36 schrieb Rod Widdowson: > +1 from me (as the source of the problem). > > What's the story with the projects\vcXXX tree. I think that it is less > capable than the command line, but we are trying to move our projects that > way with, "msbuild foo.sln /p:Debug|Release

curl WinSSL and libssh2 WinCNG auto-build downloads

Hello everyone, I am now automatically building a release (debug features disabled) version of curl and libssh2 after running the test suite for my WinSSL/SChannel (Windows-native TLS-backend) auto-builds. You can get the full build directory after buildconf, configure and make from the following

Re: Build libcurl with NSS support

Hello everyone, > Am 13.06.2016 um 20:22 schrieb Cody Pritchard (cpritchard) > : > > But, towards the end it lists a bunch of warnings about linking errors to the > NSS libraries stating: > Warning: linker path does not have real file for library -lssl3. > I

Running the test suite in a self-cleaning Windows process job group

Hello everyone, I wrote a small utility program called group-job [1] to run the test suite in a Windows process job group that automatically causes all processes to be killed if the main process ends or dies. Example usage within a Windows continuous integration environment: group-job.exe

[PATCH] tool_operate.c: add Windows support for file-uploads using stdin

-Windows implementation in lib/telnet.c also does not handle end-of-file if a custom read function is used. Best regards, Marc From 86cdc842cf0b99dc2e4484f2e6ddebfc01961aec Mon Sep 17 00:00:00 2001 From: Marc Hoersken <i...@marc-hoersken.de> Date: Wed, 16 Dec 2015 16:23:59 +0100 Subject:

Re: Schannel and buffer sizes

Hey Warren, I just posted the following patch to the Github repository: https://github.com/bagder/curl/commit/212e3e26bc711c4bd2e0a1c2bbc14b4e40506917 Please test if this solves the increasing memory issue for you. Best regards, Marc

smb.c:320 warning C6297: Arithmetic overflow

Hey everyone, would someone who has been working on SMB mind taking a look at the following warning for line 320 in smb.c while building for Windows 64-bit using VS2012: warning C6297: Arithmetic overflow: 32-bit value is shifted, then cast to 64-bit value. Result may not be an expected value

Re: smb.c:320 warning C6297: Arithmetic overflow

On 14.12.2014 20:33 Steve Holme wrote: I don't get the warning myself on VS2012 with warning level /W3 or when extending it to /W4 or /Wall so I'm a little blind here :( Is there anything I can enable in the generated project files to see what you're seeing? I used the code analysis

Re: smb.c:320 warning C6297: Arithmetic overflow

On 14.12.2014 21:00 Steve Holme wrote: I read the MSDN article for that warning and believe we have to cast to a size_t before the shift happens. As such, I have performed some testing on large files to make sure I don't break anything. I separated the length out into its own variable so I

Re: Memory leak in libcurl's curl_schannel.c (maybe)

Hi Melissa, On 01.10.2014 09:08 Melissa wrote: In our project, we're getting memory leak reports from our crash report system at program shutdown of our Windows builds. The leak is that connssl-cred in curl_schannel.c is not being freed. thanks for your bug report! I just pushed a patch

Re: Files in repo with mixed line endings

On 12.09.2014 um 10:25 Daniel Stenberg wrote: On Thu, 11 Sep 2014, Ray Satiro wrote: Attached is a patch which changes all CRLF line endings to LF in the three scripts with mixed line endings. Thanks! Diffs are really bad at handling newlines so this patch doesn't really apply easily - but

Re: schannel modifications for WinCE 6

On 11.09.2014 um 14:57 Ben Sutcliffe wrote: - Fix an apparent bug in hostname verification for wildcard certs. For *. example.com http://example.com from the cert, it was comparing .example.com http://example.com instead of example.com http://example.com

Re: better random from TLS/crypto backends!

On 30.07.2014 00:12, Daniel Stenberg wrote: So plese, help us out and make sure libcurl still build with your favourite TLS backend by making use of its suitable random function! Steve, Pierre and myself already proposed a possible solution for the WinSSL / SChannel backend, please see:

Re: NSS builds (was Re: [Survey] What people want us to do next)

On 16.06.2014 09:49, Daniel Stenberg wrote: We do NSS builds on Linux pretty frequently so we do keep that build in shape. We need volunteers to get the NSS stuff tested and the build polished on Windows. I had a Windows NSS build quite some time in order to build it for use with Pidgin

Re: NSS builds (was Re: [Survey] What people want us to do next)

On 16.06.2014 12:17, Marc Hoersken wrote: I had a Windows NSS build quite some time in order to build it for use with Pidgin plugins. I will try to revive it for public use. :-) Could someone point me to or provide up-to-date instructions on how to cross-compile nss and nspr on Debian Linux

Re: [Survey] What people want us to do next

Daniel, thanks for gathering the feedback! On 14.06.2014 23:16, Daniel Stenberg wrote: up-to-date windows binaries by knauf If I would know which type of Windows builds are the most frequently requested ones, I could setup my Windows testing buildbots [1] to produce curl and libcurl binaries

Re: weak randomness with some TLS backends

On 04.06.2014 14:55, Pierre Joye wrote: I think it is time to drop windows 2000. Maybe even xp at some point. But this function is available for xp, so it is not a issue at this point. I also suggest dropping support for Windows 2000 and Windows XP before Service Pack 3 if we want to have

Re: weak randomness with some TLS backends

On 14.06.2014 23:56, Marc Hoersken wrote: An alternative to loading Advapi32.dll dymically using LoadLibrary could be to statically link with Advapi32.lib which takes care of it. This could also be an approach for other areas in libcurl which do currently use LoadLibrary, for example: lib

Re: Autobuild failures

On 07.05.2014 22:04, Steve Holme wrote: * GetFileSizeEx() as normally declared in winbase.h and used in /tests/server/sockfilt.c doesn't appear to be defined [1] [1] - http://curl.haxx.se/dev/log.cgi?id=20140507164005-11370#prob1 I tested GetFileSizeEx() sucessfully using the five

Re: Visual Studio Project Files

Hi Steve, why don't you add that generate.bat to the repository? Even if you later update the Makefile to generate the project files for distribution, it could be useful to have it in the repository for the time being. Best regards, Marc

Re: [PATCH] tool_operate: set stdin to binary and fix telnet test 1326 on Windows

To elaborate a little bit on the actual issue on Windows: If the input is read from stdin, the code in telnet_do (telnet.c) requires that the remote socket is closed in order to terminate, because there is no way for the read callback to gracefully signal that there will be no more data to send.

Re: Visual Studio Project Files

Hi Steve, Am 18.04.2014 11:53, schrieb Steve Holme: Added VC11 files and an updated generate.bat attached for anyone wanting to test the current repo. thanks for your great efforts and especially the VC11 project files. I just used them sucessfully with Visual Studio Premium 2012. I noticed

[PATCH] tool_operate: set stdin to binary and fix telnet test 1326 on Windows

f86b2d7694a83fbe3994822bd2b45bbe64fdbd0a Mon Sep 17 00:00:00 2001 From: Marc Hoersken i...@marc-hoersken.de Date: Sat, 5 Apr 2014 21:51:39 +0200 Subject: [PATCH] tool_operate: set stdin to binary and fix telnet test 1326 on Windows --- src/tool_operate.c | 4 1 file changed, 4 insertions(+) diff --git a/src

Dead result assignments within tool param/cfg handling

Hello everyone, there seem to be dead result assignments within the tool param/cfg handling if there is a memory allocation error. Would someone who is familiar with the recent changes to the param/cfg handling logic mind taking a look? http://curl.haxx.se/scans/2014-03-22-1/ Best regards, Marc

Re: Dead result assignments within tool param/cfg handling

On 22.03.2014 18:04, Steve Holme wrote: I've just pushed a couple of fixes for my oversight - thanks again for pointing them out. Thanks for fixing them. I had similar changes on my mind, but wasn't sure whether memory allocation errors should just be skipped for individual options or abort the

Re: Filtering the autobuilds page

On 21.03.2014 09:52, Daniel Stenberg wrote: On Fri, 21 Mar 2014, Dan Fandrich wrote: I'm attaching a static example of today's autobuilds page for reference. It's not perfect, but it was quick and should be handy. If there aren't any objections, I'll add it to the site. Looks all good and

Re: [curl] Don't omit CN verification in SChannel when an IP address is used. (#94)

Hello everyone, I have just merged and pushed a slightly modified version of David's pull request to the main repository. David, thanks for spotting this. Since the change has some side-effects as SChannel and the CryptoAPI are not fully compliant with RFC 2818 section 3.1, I added the following

Re: weak cipher suites with OpenSSL, SecureTransport and... ?

Am 13.02.2014 08:27, schrieb Daniel Stenberg: Go for it. We've tried to make this happen with the other SSL backends so it makes perfect sense to me! After pushing the change on the 31th of January, I did now notice that there seems to be a problem with stunnel and Schannel while using TLSv1.2.

Re: weak cipher suites with OpenSSL, SecureTransport and... ?

Am 22.02.2014 16:04, schrieb Marc Hoersken: After pushing the change on the 31th of January, I did now notice that there seems to be a problem with stunnel and Schannel while using TLSv1.2. Disabling it and only allowing SSLv3, TLSv1.0 and TLSv1.1 on either site (stunnel config or Internet

Re: [bagder/curl] 623914: warnless: add wrapper function for read and write on Windows

Hi Steve, Am 16.02.2014 12:33, schrieb Steve Holme: Thank you - that's been bugging me for some time with my Win64 builds but I never thought of using curlx functions to solve it ;-) You are welcome. Thanks for fixing the copyright years of my commits again and again. I will try to remember

Re: More testsuite line-ending improvements to support Windows

Am 04.02.2014 00:29, schrieb Daniel Stenberg: Test 1316 may be a tricky case since it is FTP over HTTP, which basically is initially FTP but it becomes HTTP when it uses an HTTP proxy to get the job done. I do actually think that curl does not handle the HTTP header correctly in this case. It

Re: More testsuite line-ending improvements to support Windows

Hello everyone, I just rebased the 5 previously sent patches and ran the test suite on Debian 6.0.8. Test case 1316 aside, the results look promising: TESTDONE: 890 tests out of 891 reported OK: 99% TESTFAIL: These test cases failed: 1316 TESTDONE: 926 tests were considered during 468 seconds.

Re: More testsuite line-ending improvements to support Windows

Am 02.02.2014 11:43, schrieb Steve Holme: Just a quick email from FOSDEM so I will take a look at this email and your patches in further detail later... but have you noticed that on LF based systems that test 1316 is now failing [1], so I just wanted to let you know in case you're not

Re: More testsuite line-ending improvements to support Windows

:00:00 2001 From: Marc Hoersken i...@marc-hoersken.de Date: Sat, 1 Feb 2014 14:02:34 +0100 Subject: [PATCH 4/5] curl: output protocol headers using binary mode Since protocol headers contain explicit line-endings there should be no automatic conversion to ASCII text or CRLF line-endings. This might

More testsuite line-ending improvements to support Windows

: Marc Hoersken i...@marc-hoersken.de Date: Sat, 1 Feb 2014 01:01:10 +0100 Subject: [PATCH 3/3] test 500: workaround low timer resolution on Windows Since the timer resolution is lower, there are actually cases that the compared values are equal. Therefore we check for previous timestamps being

Re: BUG: NULL pointer dereference in Curl_ssl_getsessionid.

Thanks for the clarification, Daniel. From what I understood I think it would make sense to copy the SSL session information from the original easy handle into the spare closure handle, just like Vuurvlieg suggested. The SSL session information may be considered part of the connection state since

Re: BUG: NULL pointer dereference in Curl_ssl_getsessionid.

2013/5/7 Daniel Stenberg dan...@haxx.se: Just to clarify: we can close #1219 now then? https://sourceforge.net/p/curl/bugs/1219/ Yes, since my commit fixes the invalid memory access which leads to the crash. --- List admin:

Re: Proposed changes to SSL comparison documentation

2013/5/4 Nick Zitzmann n...@chronosnet.com: The documentation at http://curl.haxx.se/docs/ssl-compared.html is missing sections for Windows- and Darwin-native SSL, and also doesn't mention a few key differences between engines, like whether they're database-driven or file-driven or both, or

Re: BUG: NULL pointer dereference in Curl_ssl_getsessionid.

2013/5/4 Marc Hoersken i...@marc-hoersken.de: I just found a new way to workaround this problem with the state being replaced. The attached patch introduces a new field in the internal session structure to keep track of the currently cached session. With this change the invalid memory

Re: BUG: NULL pointer dereference in Curl_ssl_getsessionid.

2013/4/29 Daniel Stenberg dan...@haxx.se Could be, sure. But I gave this a glance and it wasn't obvious to me where the problem is... I was looking into this issue again, but I still cannot understand why data-state.session is NULL in Curl_ssl_getsessionid during Curl_ssl_shutdown. gdb.log

Re: BUG: NULL pointer dereference in Curl_ssl_getsessionid.

I just found a new way to workaround this problem with the state being replaced. The attached patch introduces a new field in the internal session structure to keep track of the currently cached session. With this change the invalid memory access is gone and the Schannel Credential handle

Re: BUG: NULL pointer dereference in Curl_ssl_getsessionid.

2013/5/4 Vuurvlieg vuurvl...@gmail.com: It is because of the line @ multi.c:1767 conn-data = multi-closure_handle; It replaces the correct state (which has session properly allocated) with the closure_handle which does not have it. I can't figure out the correct fix. Should it be illegal

Re: BUG: NULL pointer dereference in Curl_ssl_getsessionid.

Hi Daniel, 2013/4/20 Daniel Stenberg dan...@haxx.se: Any chance one of you can get some debug info there so that we get to know on what exact source line it crashes and some variable/argument contents? Here are the details you asked for: (gdb) run Starting program: d:\OS\curl\src\curl.exe -g

Re: BUG: NULL pointer dereference in Curl_ssl_getsessionid.

2013/4/18 Vuurvlieg vuurvl...@gmail.com: This results in a curl binary with the crash reported in the first post. When testing your binary did you make sure you used https protocol? Yes, of course. I used the curl command that you posted. I also used the exact same build steps on Windows 7 SP1

Re: BUG: NULL pointer dereference in Curl_ssl_getsessionid.

2013/4/19 Michael Wood esiot...@gmail.com: Hi On 18 April 2013 19:56, Vuurvlieg vuurvl...@gmail.com wrote: [...] I build it like this in the MinGW shell: $ ./buildconf buildconf: autoconf version 2.68 (BAD) Unpatched version generates unusable configure script. [...] Is this

Re: BUG: NULL pointer dereference in Curl_ssl_getsessionid.

Hi there, 2013/4/18 Vuurvli3g vuurvl...@gmail.com I recently started using libcurl and compiled it with MinGW. I wanted to use the windows built-in certs so I opted for using WinSSL. This combination crashes. Can easily be reproduced: just run curl.exe -g https://google.com The version

Re: Test 1900 isn't working due to a Linux-specific call

Hi there, 2013/4/16 Linus Nielsen Feltzing li...@haxx.se: Please try the attached patch. It uses poll() instead of epoll(). since I am currently trying to make the test suite work on Windows, it would be great if we could avoid those Unix-only functions. Neither epoll nor poll are available on

[PATCH ]Cookie sort order not deterministic

Hello everyone, I noticed that the cookie sort order is not deterministic between platforms if there are multiple cookies with the same path (length). This makes testcase 62 fail on Windows. In order to make the cookie sort function more deterministic, I would like to propose a fallback-based

Re: [bagder/curl] e51b23: getpart.pm: Strip carriage returns to fix Windows ...

Hi Daniel, 2013/4/7 Daniel Stenberg dan...@haxx.se: I propose that we introduce a new attribute for data sections in the test files for these, so that in addition to what my first patch draft from last night (that adds the ability to provide the direcory listing in the test file), marking

State of NTLM test cases on Windows

Hello everyone, I would like to note that the majority of NTLM test case failures on Windows are caused by the fact that the Negotiate Security Support Provider creates NTLM tokens that are a little different than these created by the GSSAPI. The main reason is that Windows appends an 8 byte OS

Re: State of NTLM test cases on Windows

2013/4/7 Marc Hoersken i...@marc-hoersken.de: Another issue is that the NTLM type 1 mesage flags are set differently. Example follow up regarding the different flags: Windows supports 128-bit encryption while using NTLM authentication, but the test cases respond with the 128-bit encryption flag

Re: [PATCH ]Cookie sort order not deterministic

2013/4/7 Daniel Stenberg dan...@haxx.se: I edited the patch slightly, see attachment - bascially I removed two compiler warnings (see below) and the 'cmp' variable. I'll push this version soon unless someone objects before that. Looks good, please go ahead.

Re: [bagder/curl] efa548: ftp.pm: Made Perl testsuite able to kill Windows p...

Hi Gisle, 2013/4/6 Gisle Vanem gva...@broadpark.no: ftp.pm: Made Perl testsuite able to kill Windows processes This patch uses a tool from Windows-XP Professional: system(taskkill $filter nul 21); I run Windows-XP Home and have no such tool. AFAICS, one cannot get this from MS w/o

Re: [bagder/curl] efa548: ftp.pm: Made Perl testsuite able to kill Windows p...

Hi LRN, 2013/4/6 LRN lrn1...@gmail.com: You can use Win32API Perl module and TerminateProcess or SafeTerminateProcess. Also, CreateJobObject + AssignProcessToJobObject can be used to implement behaviour similar to kill -pid. Obviously, that also requires Win32API module. I added support for

Re: [bagder/curl] efa548: ftp.pm: Made Perl testsuite able to kill Windows p...

Hi LRN, 2013/4/6 LRN lrn1...@gmail.com: The only advice i can give is to treat tskill failures gracefully. W32 is known to put processes into a state in which they still exist, but cannot be interacted with, probably due to the fact that OS is dismantling them at the moment. Killing such

Re: [bagder/curl] e51b23: getpart.pm: Strip carriage returns to fix Windows ...

2013/4/6 Daniel Stenberg dan...@haxx.se: Note that even on non-windows we check for CRLF line endings in places, so possibly such unconditional code will remove some level of checks in tests now on msys/mingw... Do you think we should add an attribute to the test data files to specify that the

Re: [curl:bugs] #1187 winsspi ssl engine was unable connect through ssl (fwd)

Hello everyone, 2013/2/4 Daniel Stenberg dan...@haxx.se I need a little help from someone with decent Windows knowledge to comment on and massage this issue! Daniel, thanks for the separate hint on this issue. Otherwise I probably would have overlooked it. winsspi ssl engine was unable

Re: [bagder/curl] 13606b: build: make use of 93 lib/*.c renamed files

Hi everyone, 2013/1/4 Yang Tse yangs...@gmail.com: Hi, I've re-read all messages in this thread a couple of times, if you wish to reply to this message, please, don't do it until you've read it fully ... thanks for this summary, Yang. Before pushing this reversion, I'm here asking you all

Re: [bagder/curl] 13606b: build: make use of 93 lib/*.c renamed files

2013/1/3 Marc Hoersken i...@marc-hoersken.de: I haven't tested this myself, but from my experience with previous renames, GitHub and all the other tools in the git ecosystem will show you the history till the last rename, but won't automatically follow the history before that point. I have

Re: [bagder/curl] 13606b: build: make use of 93 lib/*.c renamed files

2013/1/3 Marc Hoersken i...@marc-hoersken.de: 2013/1/3 Marc Hoersken i...@marc-hoersken.de: I haven't tested this myself, but from my experience with previous renames, GitHub and all the other tools in the git ecosystem will show you the history till the last rename, but won't automatically

Re: [bagder/curl] 13606b: build: make use of 93 lib/*.c renamed files

2013/1/3 Kamil Dudka kdu...@redhat.com: I haven't tested this myself, but from my experience with previous renames, GitHub and all the other tools in the git ecosystem will show you the history till the last rename, but won't automatically follow the history before that point. I am not sure

Re: test suite on windows (was Re: HTTPS via schannel not working with multi-interface)

Hello everyone, 2012/7/11 Guenter li...@gknw.net: ok, I didnt look carefully anough - the Winsock initialisation is already there in sockfilt.c; so remains this one which needs to be fixed: #ifdef USE_WINSOCK /* ** WinSock select() does not support standard file descriptors, ** it can

Re: The Most Dangerous Code in the World

Hi, 2012/11/4 Oscar Koeroo okoe...@nikhef.nl lib/curl_schannel.c VerifyHost is not used I don't think this is correct. Please see line 163. It's RFC2818 compliant out of the box, like NSS. It's the only SSL security opt-out SSL interface I've seen. To switch it off you'll need to set the

Re: The Most Dangerous Code in the World

Hi, 2012/11/4 Oscar Koeroo okoe...@nikhef.nl: It's RFC2818 compliant out of the box, like NSS. It's the only SSL security opt-out SSL interface I've seen. To switch it off you'll need to set the flag SCH_CRED_NO_SERVERNAME_CHECK according to

Re: Making 7.27 problem

Hi Jeff, 2012/10/8 Jeff McKay jeff.mc...@comaxis.com: I tried a quick test of 7.27/WinSSL using my existing application. I recompiled it using the new import library and headers, and substituted the new libcurl.dll. No changes to my coding. Initial results don't look too promising. I

Re: Making 7.27 problem

Hi there, 2012/10/7 Jeff McKay jeff.mc...@comaxis.com: Yes, this seems to work. I changed line 109 in MakefileBuild.vc to be: SSL_CFLAGS = /DUSE_SSLEAY /DUSE_OPENSSL /I$(DEVEL_INCLUDE)/openssl could you please give the attached patch a try, too? Since ssluse.c is designed to work without

Re: Making 7.27 problem

Hi, 2012/10/6 Jeff McKay jeff.mc...@comaxis.com: nmake /f makefile.vc mode=dll VC=10 WITH_SSL=dll DEBUG=no ENABLE_IDN=no GEN_PDB=no ENABLE_SSSPI=yes I get an immediate compile error on line 317 of urldata.h, saying: 'connecting_state' struct member redefinition. From what I can tell

Re: curl (WinSSL) fails with memory buffer error (fwd)

Hello everyone, I had to push a follow up patch as fbf3560886. The download was still failing for me sometimes since the server seems to return very big encrypted data packages and Schannel was unable to decrypt the data partially. Therefore I have completely remove the internal buffer limit and

Re: Making 7.27 problem

2012/10/6 Jeff McKay jeff.mc...@comaxis.com: Thank you - that gets me a bit further, now I am getting compile errors on ssluse.c. It cannot find MD5 definitions, e.g.: 'MD5_CTX' : undeclared identifier. My current make command is: Please post the whole log output as an attachment. Otherwise

Re: Making 7.27 problem

2012/10/7 Jeff McKay jeff.mc...@comaxis.com: Sure, here it is, attached (sending to you separately, in case the list does not accept attachments). The list accepts attachments, but thanks anyway. And please stop top posting. ;-) I see MD5_CTX defined in md5.c, but I don't see that being

Re: curl (WinSSL) fails with memory buffer error (fwd)

Hi again, 2012/9/28 Marc Hoersken i...@marc-hoersken.de: I think the problem is clear: The buffer size limit of the schannel implementation is reached. This is caused by the fact that curl wants to read more data than what is available in the decrypted data buffer which then leads to more

Re: Array Index size_t vs int

Hi Daniel, 2012/9/22 Daniel Stenberg dan...@haxx.se: What I'll do next: Change the script to only run once per hour. Follow-up and make the script to clear the previous output until the script has completed. Finally I'll work on fixing it so that it doesn't re-parse all output on every

Re: Array Index size_t vs int

Hi Daniel, 2012/9/22 Daniel Stenberg dan...@haxx.se: I don't have any problems with using (s)size_t as index in arrays. I'm talking about that the conversion from those types to int that may exist in the code, are sometimes present because of inconsistent APIs in which older systems don't

Array Index size_t vs int

Hello everyone, I noticed that quite some warnings in the curl source code are related to size_t/ssite_t to int conversions. This is caused by the fact that we are mostly using int-typed indexes to access array fields there the actual array holds a string- or data-buffer. Many of the functions

Streamlining plattform checks

Hello everyone, while looking over some source code and the CONTRIBUTE document I noticed that the following statements tends not to be followed through the codebase: 2.7 Platform Dependent Code Use #ifdef HAVE_FEATURE to do conditional code. We avoid checking for particular operating

Re: [PATCH] socks.c: Added support for IPv6 connections through

Hi Daniel, 2012/9/14 Daniel Stenberg dan...@haxx.se Thanks! Looks clean and simple enough to just proceed with I think. I consider this a bug fix. Thanks for looking into the patches. I just pushed them into the repository. Best regards, Marc

Re: [PATCH] tool_metalink.c: Added support for Microsoft Windows CryptoAPI

Hello everyone, thanks to Guenter for testing the implementation on Windows XP. I just pushed the changes to the git repository. Best regards, Marc --- List admin: http://cool.haxx.se/list/listinfo/curl-library Etiquette:

[PATCH] socks.c: Added support for IPv6 connections through

Hello everyone, I just finished my implementation of IPv6 connections through SOCKSv5 proxies. Attached you will find two patches, one with the implementation itself and another one with added support inside the error messages. I would appreciate any feedback on these patches and would like to

[PATCH] tool_metalink.c: Added support for Microsoft Windows CryptoAPI

Hello everyone, Anthony approached me some weeks ago and asked if it is possible to make use of built-in Windows functions in order to provide the hashing functionality required by Metalink. Yes, it is entirely possible with the built-in CryptoAPI and I just finished a working implementation. I

Re: bug in schannel connection shutdown?

Hello everyone, 2012/8/31 Daniel Stenberg dan...@haxx.se On Thu, 9 Aug 2012, Salisbury, Mark wrote: If I then call curl_multi_cleanup() (when shutting down the entire program), I get accesses to free()d memory in schannel connection cleanup. I don't get such issues on linux with gnutls. The

Re: bug in schannel connection shutdown?

Hello again, I was able to reproduce the issue and made sure that the patch fixes the IMA/SEGFAULT. The slightly adjusted changes were just pushed. Best regards, Marc --- List admin: http://cool.haxx.se/list/listinfo/curl-library

Re: [PATCH] winbuild clean up and alignment changes

Hello again, since I didn't receive neither positive feedback nor negative issues regarding these patches, I am going to push them in a few minutes. Best regards, Marc --- List admin: http://cool.haxx.se/list/listinfo/curl-library

Re: [STEP 3] SSPI/Schannel string literals voting process

[Version_Literal=WinSSL] [Config_Literal=winssl] 2012/7/4 Yang Tse yangs...@gmail.com Hi friends, [STEP 2] of this process has resulted in the following candidate lists. Two candidates for the [STEP 3] Version_Literal voting: [Version_Literal=Schannel] [Version_Literal=WinSSL] Two

Re: Makefile.m32: Run del and copy commands through cmd

Hi Guenter, 2012/7/3 Guenter li...@gknw.net: @Marc: can you please test if the recent makefiles still fail for you with Eclipse? If so can you then pls test if $(shell ...) can fix it? Thanks for the changes, they are working just fine. It seems to use rm for file removal now, but building and

Re: Unicode and NTLM

Hi there, 2012/7/2 Daniel Stenberg dan...@haxx.se Thanks, pushed now with some minor edits by me. the changes introduced a warning about the unused variable passwd. Attached you will find a patch to remove that variable. Best regards, Marc

Re: gitignore: Added files generated by mingw32, eclipse and VC

Hi Günter, 2012/7/3 Guenter li...@gknw.net: just looked at it: --- a/.gitignore +++ b/.gitignore @@ -42,3 +42,12 @@ TAGS *~ aclocal.m4.bak CHANGES.dist +.project +.cproject +lib/libcurl.a +lib/libcurl.res +lib/libcurldll.a +src/curl.res +winbuild/CURL_OBJS.inc

gitignore: Added files generated by mingw32, eclipse and VC

Hello everyone, I updated the .gitignore file to exclude files generated by mingw32, eclipse and VC. Would be great if someone could push it to the main repo. Thanks in advance! Best regards, Marc 0001-gitignore-Added-files-generated-by-mingw32-eclipse-a.patch Description: Binary data

Makefile.m32: Run del and copy commands through cmd

Hello everyone, here is a follow-up patch to my previous .gitignore patch. It makes it possible to build and clean curl using the eclipse CDT MinGW compatible compiler tool chain. It introduces one small regression: The del command will now show errors/warnings if a file cannot be deleted since

Re: warning: ISO C does not support the 'I64' ms_printf length modifier

Hi Guenter, 2012/6/30 Guenter li...@gknw.net: so my build has the additional flag -Wno-pedantic-ms-format - my guess is that this might fix the warnings for your build too, and that your compiler is not correctly detected as mingw in configure, probably due to no correct triplet

warning: ISO C does not support the 'I64' ms_printf length modifier

Hello everyone, I am getting the following warning from my mingw32 builds:   warning: ISO C does not support the 'I64' ms_printf length modifier For example: http://curl.haxx.se/dev/log.cgi?id=20120629151115-27110#prob1 This is the corresponding line: fprintf(stderr, *** We read %

Re: One more proposal: Getting a trust

Hi there, 2012/6/28 Nick Zitzmann n...@chronosnet.com: Let me know what you think… Is this something the Windows SSL support could use as well? I can't think of any usage for Windows SSL. Schannel does not provide such a trust handle since it returns which certification or SSL connection

Re: warning: ISO C does not support the 'I64' ms_printf length modifier

Hi Guenter, 2012/6/30 Guenter li...@gknw.net: Hi Marc, [...] I think this is a bug with the MinGW compilers, and not fixable; I tried a lot of stuff, did google for hours, but all I could change was the error message; these printf identifiers are legal for Windows, and same as what MinGW

Re: warning: ISO C does not support the 'I64' ms_printf length modifier

Hi Guenter, 2012/6/30 Guenter li...@gknw.net: IIRC there's also an issue with length specifiers, f.e. '%I64d' gives another warning than '%8I64d' ... but regardless of the gcc warnings: all these printf formats finally work correctly ... I believe this issue should be entered as bug at the

Re: SSPI/Schannel Config_Literal voting rules

Hi everyone, 2012/6/28 Yang Tse yangs...@gmail.com: Yes, yes. I'm certainly in favour of that. Although I want explicit approval on this thread from Marc Hoersken in order to interpret his proposal in this way, given that his proposal would be affected by this. Of course I would like

Re: SSPI/Schannel Config_Literal voting rules

2012/6/28 Michael Wood esiot...@gmail.com: On 28 June 2012 14:06, Marc Hoersken i...@marc-hoersken.de wrote: [Config_Literal=WINDOWS_SSL] configure: --with-windows-ssl or --with-winssl (short form, corresponding to other parameters) other: WITH_WINDOWS_SSL or ENABLE_WINDOWS_SSL I'm not sure

Re: [PATCH] ssluse.c: Fixed compile error: conversion to 'int' from 'size_t'

Hello everyone, attached you will find two patches to fix the warnings I reported earlier. Would be great if someone could do a second test and push afterwards. Best regards, Marc 0001-nss-Fixed-size_t-conversion-warnings.patch Description: Binary data

Re: [PATCH] ssluse.c: Fixed compile error: conversion to 'int' from 'size_t'

I've already seen your previous post from 20 hours ago. Please be patient it will be addressed shortly. I am patient... I just thought that it would be a good idea to do the patching myself.. --- List admin:

Re: [STEP 2] SSPI/Schannel string literals voting process

My four votes: [Version_Literal=Schannel] [Version_Literal=WinSSL] [Config_Literal=schannel] [Config_Literal=winssl] --- List admin: http://cool.haxx.se/list/listinfo/curl-library Etiquette: http://curl.haxx.se/mail/etiquette.html

Re: Peer review! SecureTransport (native SSL on iOS/OS X) patch

Thanks for incorporating my suggestions and the patch! 2012/6/26 Nick Zitzmann n...@chronosnet.com: - In st_connect_step1 replace multiple calls to SSLSetProtocolVersionEnabled with an internal variable and just one call to SSLSetProtocolVersionEnabled. That is unfortunately a necessary

Re: [PATCH] ssluse.c: Fixed compile error: conversion to 'int' from 'size_t'

Hi Yang, 2012/6/27 Yang Tse yangs...@gmail.com I've fixed it diferently. Fix already pushed to repo. could you apply the same fix to the following warnings, please? goto problem 2 ../.././lib/nss.c:1541: warning: conversion to 'int' from 'size_t' may alter its value ../.././lib/nss.c: In

  1   2   >