Re: [PATCH] rawstr: Speed up Curl_raw_toupper by 40%

Hi Lauri, your code wouldn't work with non-continuous character sets, like EBCDIC. Did you turn on optimization when compiling curl/libcurl ? Any half-way modern compiler should translate the current code into a lookup table or code like yours. Another option would be to check for a

Re: [PATCH] rawstr: Speed up Curl_raw_toupper by 40%

On Mon, 02 Nov 2015 16:39:55 +0100 Tim Ruehsen wrote: > Hi Lauri, > > your code wouldn't work with non-continuous character sets, like EBCDIC. > > Did you turn on optimization when compiling curl/libcurl ? Any half-way > modern > compiler should translate the current code

RE: Is the curl library what I need?

Hi Daniel, > With that little conditional in mind, I would say that libcurl will > significantly simplify your effort as you'll get fully standards compliant > HTTPS capable POSTs (and more) with no basically no work more than > building it for your platform. As this is a BIOS extension space is

Re: [PATCH] formadd: support >2GB files on windows

On Fri, 30 Oct 2015, Daniel Stenberg wrote: The original problem being that curl_formadd() right now uses 'long' for a size so that it can't be set larger than 2GB on any windows version (even if 64bit) and other 32bit systems. This patch introduces a new option to pass in a 'curl_off_t'

Is the curl library what I need?

Hello, I am part way through a project developing an x86 BIOS extension that needs to communicate with and transfer files from an Apache web server initially using http. The network stack within the BIOS extension is lwIP and in the future it may also include an SSL/TLS library such as WolfSSL

[PATCH] rawstr: Speed up Curl_raw_toupper by 40%

Rationale: when starting up a curl-using app, all cookies from the jar are checked against each other. This was causing a two-second startup delay in the Fifth browser. All tests pass. Signed-off-by: Lauri Kasanen --- lib/rawstr.c | 57

Re: [PATCH] rawstr: Speed up Curl_raw_toupper by 40%

On Mon, Nov 02, 2015 at 05:09:45PM +0200, Lauri Kasanen wrote: > + if (in >= 'a' && in <= 'z') > +return 'A' + in - 'a'; > + Doesn't this change assume ASCII? Frank -- "Debugging is twice as hard as writing the code in the first place. Therefore, if you write the code as cleverly as

[PATCH v2] rawstr: Speed up Curl_raw_toupper by 40%

Rationale: when starting up a curl-using app, all cookies from the jar are checked against each other. This was causing a two-second startup delay in the Fifth browser. All tests pass. Signed-off-by: Lauri Kasanen --- v2: Guards as suggested by Tim. The speedup still correctly

Re: [PATCH] rawstr: Speed up Curl_raw_toupper by 40%

On Monday 02 November 2015 17:49:09 Lauri Kasanen wrote: > On Mon, 02 Nov 2015 16:39:55 +0100 > > Tim Ruehsen wrote: > > Hi Lauri, > > > > your code wouldn't work with non-continuous character sets, like EBCDIC. > > > > Did you turn on optimization when compiling

RE: [PATCH] rawstr: Speed up Curl_raw_toupper by 40%

Lauri Kasanen wrote: > All comparisons were made with -O3 -march=native. Does curl really support ebcdic? In 2015? Yes, curl supports EBCDIC: it is used for the TPF platform on IBM mainframe. iSeries/OS400 implementation runs internally in ASCII, although the machine is EBCDIC-based. And

Re: [PATCH v2] rawstr: Speed up Curl_raw_toupper by 40%

On Mon, 2 Nov 2015, Lauri Kasanen wrote: +#if ('z' - 'a' == 25) && ('Z' - 'A' == 25) I have one concern with this method: it checks that the compiler is using ascii, not necessarily that the target host is using it. I'm not sure the distinction actually matters. However, I also think that

[PATCH v3] rawstr: Speed up Curl_raw_toupper by 40%

Rationale: when starting up a curl-using app, all cookies from the jar are checked against each other. This was causing a two-second startup delay in the Fifth browser. All tests pass. Signed-off-by: Lauri Kasanen --- v3: Test with CURL_DOES_CONVERSIONS. lib/rawstr.c | 6 ++

Re: [PATCH] rawstr: Speed up Curl_raw_toupper by 40%

Just an idea. At least check #if ('z' - 'a' == 25) && ('z' - 'a' == 25) ... your code #else ... old code #endif (didn't test it) Tim On Monday 02 November 2015 17:49:09 Lauri Kasanen wrote: > On Mon, 02 Nov 2015 16:39:55 +0100 > > Tim Ruehsen wrote: > > Hi Lauri, > > > >

Threaded resolver is not enabled on Solaris

Hi All, I tried to enable threaded resolver on Solaris by using the option "--enable-threaded-resolver" but when "configure" completes it show that the resolver is "default". I used the same option on Linux without any issues. Following is the error that I get in the "configure" step:

Re: Threaded resolver is not enabled on Solaris

On Mon, 2 Nov 2015, Ahmed, Najeeb wrote: Please note that I was able to enable threaded resolver on Curl version 7.45.0 but I need to make it work on Curl version 7.30.0 as this is the version that we currently use in our product. Can someone please let me know what changes do I need to make

Re: [PATCH v3] rawstr: Speed up Curl_raw_toupper by 40%

On Mon, 2 Nov 2015, Lauri Kasanen wrote: Rationale: when starting up a curl-using app, all cookies from the jar are checked against each other. This was causing a two-second startup delay in the Fifth browser. Merged as commit 3bd7f2800. Thanks again! -- / daniel.haxx.se

our feature window is closing

Hey We'll close the feature window on midnight Wednesday (about 60 hours left). If you want to merge something for this pending release that isn't a plain bugfix, you better hurry up. =) After the window closes, we'll only merge bug fixes until the release day. curl 7.46.0 is targetted for

RE: Is the curl library what I need?

On Mon, 2 Nov 2015, Andy Pont wrote: As this is a BIOS extension space is at a premium, how small can the library be got? docs/INSTALL was recently updated with a refreshed number and it says 109KB on i386 for an HTTP or FTP only library with a somewhat limited feature set. -- /