cppcms-users Mailing List for CppCMS C++ Web Framework (Page 15)
Brought to you by:
artyom-beilis
Menu
▾
▴
cppcms-users — Disussions of users of CppCMS
You can subscribe to this list here.
| 2009 |
Jan
|
Feb
(22) |
Mar
|
Apr
(3) |
May
|
Jun
(4) |
Jul
|
Aug
|
Sep
|
Oct
(15) |
Nov
(16) |
Dec
(13) |
|---|---|---|---|---|---|---|---|---|---|---|---|---|
| 2010 |
Jan
(4) |
Feb
|
Mar
(8) |
Apr
(8) |
May
(8) |
Jun
(36) |
Jul
(63) |
Aug
(126) |
Sep
(47) |
Oct
(66) |
Nov
(46) |
Dec
(42) |
| 2011 |
Jan
(87) |
Feb
(24) |
Mar
(54) |
Apr
(21) |
May
(22) |
Jun
(18) |
Jul
(22) |
Aug
(101) |
Sep
(57) |
Oct
(33) |
Nov
(34) |
Dec
(66) |
| 2012 |
Jan
(64) |
Feb
(76) |
Mar
(73) |
Apr
(105) |
May
(93) |
Jun
(83) |
Jul
(84) |
Aug
(88) |
Sep
(57) |
Oct
(59) |
Nov
(35) |
Dec
(49) |
| 2013 |
Jan
(67) |
Feb
(17) |
Mar
(49) |
Apr
(64) |
May
(87) |
Jun
(64) |
Jul
(93) |
Aug
(23) |
Sep
(15) |
Oct
(16) |
Nov
(62) |
Dec
(73) |
| 2014 |
Jan
(5) |
Feb
(23) |
Mar
(21) |
Apr
(11) |
May
(1) |
Jun
(19) |
Jul
(27) |
Aug
(16) |
Sep
(5) |
Oct
(37) |
Nov
(12) |
Dec
(9) |
| 2015 |
Jan
(7) |
Feb
(7) |
Mar
(44) |
Apr
(28) |
May
(5) |
Jun
(12) |
Jul
(8) |
Aug
|
Sep
(39) |
Oct
(34) |
Nov
(30) |
Dec
(34) |
| 2016 |
Jan
(66) |
Feb
(23) |
Mar
(33) |
Apr
(15) |
May
(11) |
Jun
(15) |
Jul
(26) |
Aug
(4) |
Sep
(1) |
Oct
(30) |
Nov
(10) |
Dec
|
| 2017 |
Jan
(52) |
Feb
(9) |
Mar
(24) |
Apr
(16) |
May
(9) |
Jun
(12) |
Jul
(33) |
Aug
(8) |
Sep
|
Oct
(1) |
Nov
(2) |
Dec
(6) |
| 2018 |
Jan
(5) |
Feb
|
Mar
|
Apr
|
May
(14) |
Jun
(1) |
Jul
(9) |
Aug
(1) |
Sep
(13) |
Oct
(8) |
Nov
(2) |
Dec
(2) |
| 2019 |
Jan
(1) |
Feb
(1) |
Mar
(3) |
Apr
(3) |
May
(1) |
Jun
|
Jul
|
Aug
|
Sep
|
Oct
(2) |
Nov
|
Dec
|
| 2020 |
Jan
|
Feb
(1) |
Mar
|
Apr
|
May
|
Jun
(9) |
Jul
(6) |
Aug
(25) |
Sep
(10) |
Oct
(10) |
Nov
(6) |
Dec
|
| 2021 |
Jan
|
Feb
|
Mar
(7) |
Apr
(1) |
May
|
Jun
(1) |
Jul
|
Aug
|
Sep
(9) |
Oct
(1) |
Nov
|
Dec
|
| 2022 |
Jan
|
Feb
|
Mar
|
Apr
(3) |
May
(1) |
Jun
|
Jul
|
Aug
|
Sep
|
Oct
|
Nov
|
Dec
|
|
From: CN <cn...@fa...> - 2016-10-28 17:19:32
|
On Fri, Oct 28, 2016, at 10:21 PM, ano...@op... wrote: > Just what I want add - I am very motivated to improve CppCMS and port it > to C++11, and all what needed - new branch, where I can send patches. Hats off to you! Best Regards, CN -- http://www.fastmail.com - Email service worth paying for. Try it for free |
|
From: Marius C. <mf...@gm...> - 2016-10-28 16:02:19
|
I like Qt but I don't think it's a very good idea to use for something like this where maximum performance is the first priority. Qt is just adds extra overhead and it's just not optimized for this kind of things because it doesn't really need to be for most of the Qt use cases. On Fri, Oct 28, 2016 at 6:51 PM, sergey lavrov <ccp...@gm...> wrote: > as far as i know round-11 was done before optimisation in frog core and > thats why there are no results in round-12 and I really dont like QT in core > of frog framework. > yeah it will be very interesting to see how cppcms perfoms, because GO > results in round-12 make me cry. > > On Fri, Oct 28, 2016 at 5:50 PM, Marius Cirsta <mf...@gm...> wrote: >> >> I'm just not sure about the efficiency of treefrog. I've seen some >> basic tests that put it in a not so good light ( >> https://www.techempower.com/benchmarks/#section=data-r11&hw=peak&test=db >> ) . If a C++ framework seems to be doing so bad in performance >> compared to things like Java and is even behind LUA and PHP in some >> cases than what's the purpose of having it done in C++ ? >> I'm not sure how cppcms would perform in this test but I'm hoping it >> would perform better. >> >> On Fri, Oct 28, 2016 at 5:40 PM, sergey lavrov <ccp...@gm...> >> wrote: >> > fyi found today c++ mvc framework in active development state >> > http://www.treefrogframework.org/. development going on github (what we >> > asked artyom many times), epoll/thread, two template engines, db access, >> > better license, has all features that cppcms has or even more ... it's >> > really sad that cppcms almost died ... >> > >> > On Fri, Oct 28, 2016 at 5:21 PM, <ano...@op...> wrote: >> >> >> >> Just what I want add - I am very motivated to improve CppCMS and port >> >> it >> >> to C++11, and all what needed - new branch, where I can send patches. >> >> >> >> >> >> >> >> ------------------------------------------------------------------------------ >> >> The Command Line: Reinvented for Modern Developers >> >> Did the resurgence of CLI tooling catch you by surprise? >> >> Reconnect with the command line and become more productive. >> >> Learn the new .NET and ASP.NET CLI. Get your free copy! >> >> http://sdm.link/telerik >> >> _______________________________________________ >> >> Cppcms-users mailing list >> >> Cpp...@li... >> >> https://lists.sourceforge.net/lists/listinfo/cppcms-users >> > >> > >> > >> > >> > ------------------------------------------------------------------------------ >> > The Command Line: Reinvented for Modern Developers >> > Did the resurgence of CLI tooling catch you by surprise? >> > Reconnect with the command line and become more productive. >> > Learn the new .NET and ASP.NET CLI. Get your free copy! >> > http://sdm.link/telerik >> > _______________________________________________ >> > Cppcms-users mailing list >> > Cpp...@li... >> > https://lists.sourceforge.net/lists/listinfo/cppcms-users >> > >> >> >> ------------------------------------------------------------------------------ >> The Command Line: Reinvented for Modern Developers >> Did the resurgence of CLI tooling catch you by surprise? >> Reconnect with the command line and become more productive. >> Learn the new .NET and ASP.NET CLI. Get your free copy! >> http://sdm.link/telerik >> _______________________________________________ >> Cppcms-users mailing list >> Cpp...@li... >> https://lists.sourceforge.net/lists/listinfo/cppcms-users > > > > ------------------------------------------------------------------------------ > The Command Line: Reinvented for Modern Developers > Did the resurgence of CLI tooling catch you by surprise? > Reconnect with the command line and become more productive. > Learn the new .NET and ASP.NET CLI. Get your free copy! > http://sdm.link/telerik > _______________________________________________ > Cppcms-users mailing list > Cpp...@li... > https://lists.sourceforge.net/lists/listinfo/cppcms-users > |
|
From: Shiv S. D. <shi...@gm...> - 2016-10-28 15:54:26
|
Most of the solutions will sit behind Nginx/Apache/Lighttpd thus Lapis would benefit there as it is embedded in Nginx. On Fri, Oct 28, 2016 at 9:21 PM, sergey lavrov <ccp...@gm...> wrote: > as far as i know round-11 was done before optimisation in frog core and > thats why there are no results in round-12 and I really dont like QT in > core of frog framework. > yeah it will be very interesting to see how cppcms perfoms, because GO > results in round-12 make me cry. > > On Fri, Oct 28, 2016 at 5:50 PM, Marius Cirsta <mf...@gm...> wrote: > >> I'm just not sure about the efficiency of treefrog. I've seen some >> basic tests that put it in a not so good light ( >> https://www.techempower.com/benchmarks/#section=data-r11&hw=peak&test=db >> ) . If a C++ framework seems to be doing so bad in performance >> compared to things like Java and is even behind LUA and PHP in some >> cases than what's the purpose of having it done in C++ ? >> I'm not sure how cppcms would perform in this test but I'm hoping it >> would perform better. >> >> On Fri, Oct 28, 2016 at 5:40 PM, sergey lavrov <ccp...@gm...> >> wrote: >> > fyi found today c++ mvc framework in active development state >> > http://www.treefrogframework.org/. development going on github (what we >> > asked artyom many times), epoll/thread, two template engines, db access, >> > better license, has all features that cppcms has or even more ... it's >> > really sad that cppcms almost died ... >> > >> > On Fri, Oct 28, 2016 at 5:21 PM, <ano...@op...> wrote: >> >> >> >> Just what I want add - I am very motivated to improve CppCMS and port >> it >> >> to C++11, and all what needed - new branch, where I can send patches. >> >> >> >> >> >> ------------------------------------------------------------ >> ------------------ >> >> The Command Line: Reinvented for Modern Developers >> >> Did the resurgence of CLI tooling catch you by surprise? >> >> Reconnect with the command line and become more productive. >> >> Learn the new .NET and ASP.NET CLI. Get your free copy! >> >> http://sdm.link/telerik >> >> _______________________________________________ >> >> Cppcms-users mailing list >> >> Cpp...@li... >> >> https://lists.sourceforge.net/lists/listinfo/cppcms-users >> > >> > >> > >> > ------------------------------------------------------------ >> ------------------ >> > The Command Line: Reinvented for Modern Developers >> > Did the resurgence of CLI tooling catch you by surprise? >> > Reconnect with the command line and become more productive. >> > Learn the new .NET and ASP.NET CLI. Get your free copy! >> > http://sdm.link/telerik >> > _______________________________________________ >> > Cppcms-users mailing list >> > Cpp...@li... >> > https://lists.sourceforge.net/lists/listinfo/cppcms-users >> > >> >> ------------------------------------------------------------ >> ------------------ >> The Command Line: Reinvented for Modern Developers >> Did the resurgence of CLI tooling catch you by surprise? >> Reconnect with the command line and become more productive. >> Learn the new .NET and ASP.NET CLI. Get your free copy! >> http://sdm.link/telerik >> _______________________________________________ >> Cppcms-users mailing list >> Cpp...@li... >> https://lists.sourceforge.net/lists/listinfo/cppcms-users >> > > > ------------------------------------------------------------ > ------------------ > The Command Line: Reinvented for Modern Developers > Did the resurgence of CLI tooling catch you by surprise? > Reconnect with the command line and become more productive. > Learn the new .NET and ASP.NET CLI. Get your free copy! > http://sdm.link/telerik > _______________________________________________ > Cppcms-users mailing list > Cpp...@li... > https://lists.sourceforge.net/lists/listinfo/cppcms-users > > -- Respect, Shiv Shankar Dayal |
|
From: sergey l. <ccp...@gm...> - 2016-10-28 15:51:21
|
as far as i know round-11 was done before optimisation in frog core and thats why there are no results in round-12 and I really dont like QT in core of frog framework. yeah it will be very interesting to see how cppcms perfoms, because GO results in round-12 make me cry. On Fri, Oct 28, 2016 at 5:50 PM, Marius Cirsta <mf...@gm...> wrote: > I'm just not sure about the efficiency of treefrog. I've seen some > basic tests that put it in a not so good light ( > https://www.techempower.com/benchmarks/#section=data-r11&hw=peak&test=db > ) . If a C++ framework seems to be doing so bad in performance > compared to things like Java and is even behind LUA and PHP in some > cases than what's the purpose of having it done in C++ ? > I'm not sure how cppcms would perform in this test but I'm hoping it > would perform better. > > On Fri, Oct 28, 2016 at 5:40 PM, sergey lavrov <ccp...@gm...> > wrote: > > fyi found today c++ mvc framework in active development state > > http://www.treefrogframework.org/. development going on github (what we > > asked artyom many times), epoll/thread, two template engines, db access, > > better license, has all features that cppcms has or even more ... it's > > really sad that cppcms almost died ... > > > > On Fri, Oct 28, 2016 at 5:21 PM, <ano...@op...> wrote: > >> > >> Just what I want add - I am very motivated to improve CppCMS and port it > >> to C++11, and all what needed - new branch, where I can send patches. > >> > >> > >> ------------------------------------------------------------ > ------------------ > >> The Command Line: Reinvented for Modern Developers > >> Did the resurgence of CLI tooling catch you by surprise? > >> Reconnect with the command line and become more productive. > >> Learn the new .NET and ASP.NET CLI. Get your free copy! > >> http://sdm.link/telerik > >> _______________________________________________ > >> Cppcms-users mailing list > >> Cpp...@li... > >> https://lists.sourceforge.net/lists/listinfo/cppcms-users > > > > > > > > ------------------------------------------------------------ > ------------------ > > The Command Line: Reinvented for Modern Developers > > Did the resurgence of CLI tooling catch you by surprise? > > Reconnect with the command line and become more productive. > > Learn the new .NET and ASP.NET CLI. Get your free copy! > > http://sdm.link/telerik > > _______________________________________________ > > Cppcms-users mailing list > > Cpp...@li... > > https://lists.sourceforge.net/lists/listinfo/cppcms-users > > > > ------------------------------------------------------------ > ------------------ > The Command Line: Reinvented for Modern Developers > Did the resurgence of CLI tooling catch you by surprise? > Reconnect with the command line and become more productive. > Learn the new .NET and ASP.NET CLI. Get your free copy! > http://sdm.link/telerik > _______________________________________________ > Cppcms-users mailing list > Cpp...@li... > https://lists.sourceforge.net/lists/listinfo/cppcms-users > |
|
From: Shiv S. D. <shi...@gm...> - 2016-10-28 15:49:10
|
I have moved on to Lapis web framework. http://leafo.net/lapis/ On Fri, Oct 28, 2016 at 8:20 PM, Marius Cirsta <mf...@gm...> wrote: > I'm just not sure about the efficiency of treefrog. I've seen some > basic tests that put it in a not so good light ( > https://www.techempower.com/benchmarks/#section=data-r11&hw=peak&test=db > ) . If a C++ framework seems to be doing so bad in performance > compared to things like Java and is even behind LUA and PHP in some > cases than what's the purpose of having it done in C++ ? > I'm not sure how cppcms would perform in this test but I'm hoping it > would perform better. > > On Fri, Oct 28, 2016 at 5:40 PM, sergey lavrov <ccp...@gm...> > wrote: > > fyi found today c++ mvc framework in active development state > > http://www.treefrogframework.org/. development going on github (what we > > asked artyom many times), epoll/thread, two template engines, db access, > > better license, has all features that cppcms has or even more ... it's > > really sad that cppcms almost died ... > > > > On Fri, Oct 28, 2016 at 5:21 PM, <ano...@op...> wrote: > >> > >> Just what I want add - I am very motivated to improve CppCMS and port it > >> to C++11, and all what needed - new branch, where I can send patches. > >> > >> > >> ------------------------------------------------------------ > ------------------ > >> The Command Line: Reinvented for Modern Developers > >> Did the resurgence of CLI tooling catch you by surprise? > >> Reconnect with the command line and become more productive. > >> Learn the new .NET and ASP.NET CLI. Get your free copy! > >> http://sdm.link/telerik > >> _______________________________________________ > >> Cppcms-users mailing list > >> Cpp...@li... > >> https://lists.sourceforge.net/lists/listinfo/cppcms-users > > > > > > > > ------------------------------------------------------------ > ------------------ > > The Command Line: Reinvented for Modern Developers > > Did the resurgence of CLI tooling catch you by surprise? > > Reconnect with the command line and become more productive. > > Learn the new .NET and ASP.NET CLI. Get your free copy! > > http://sdm.link/telerik > > _______________________________________________ > > Cppcms-users mailing list > > Cpp...@li... > > https://lists.sourceforge.net/lists/listinfo/cppcms-users > > > > ------------------------------------------------------------ > ------------------ > The Command Line: Reinvented for Modern Developers > Did the resurgence of CLI tooling catch you by surprise? > Reconnect with the command line and become more productive. > Learn the new .NET and ASP.NET CLI. Get your free copy! > http://sdm.link/telerik > _______________________________________________ > Cppcms-users mailing list > Cpp...@li... > https://lists.sourceforge.net/lists/listinfo/cppcms-users > -- Respect, Shiv Shankar Dayal |
|
From: Marius C. <mf...@gm...> - 2016-10-28 14:50:34
|
I'm just not sure about the efficiency of treefrog. I've seen some basic tests that put it in a not so good light ( https://www.techempower.com/benchmarks/#section=data-r11&hw=peak&test=db ) . If a C++ framework seems to be doing so bad in performance compared to things like Java and is even behind LUA and PHP in some cases than what's the purpose of having it done in C++ ? I'm not sure how cppcms would perform in this test but I'm hoping it would perform better. On Fri, Oct 28, 2016 at 5:40 PM, sergey lavrov <ccp...@gm...> wrote: > fyi found today c++ mvc framework in active development state > http://www.treefrogframework.org/. development going on github (what we > asked artyom many times), epoll/thread, two template engines, db access, > better license, has all features that cppcms has or even more ... it's > really sad that cppcms almost died ... > > On Fri, Oct 28, 2016 at 5:21 PM, <ano...@op...> wrote: >> >> Just what I want add - I am very motivated to improve CppCMS and port it >> to C++11, and all what needed - new branch, where I can send patches. >> >> >> ------------------------------------------------------------------------------ >> The Command Line: Reinvented for Modern Developers >> Did the resurgence of CLI tooling catch you by surprise? >> Reconnect with the command line and become more productive. >> Learn the new .NET and ASP.NET CLI. Get your free copy! >> http://sdm.link/telerik >> _______________________________________________ >> Cppcms-users mailing list >> Cpp...@li... >> https://lists.sourceforge.net/lists/listinfo/cppcms-users > > > > ------------------------------------------------------------------------------ > The Command Line: Reinvented for Modern Developers > Did the resurgence of CLI tooling catch you by surprise? > Reconnect with the command line and become more productive. > Learn the new .NET and ASP.NET CLI. Get your free copy! > http://sdm.link/telerik > _______________________________________________ > Cppcms-users mailing list > Cpp...@li... > https://lists.sourceforge.net/lists/listinfo/cppcms-users > |
|
From: sergey l. <ccp...@gm...> - 2016-10-28 14:40:40
|
fyi found today c++ mvc framework in active development state http://www.treefrogframework.org/. development going on github (what we asked artyom many times), epoll/thread, two template engines, db access, better license, has all features that cppcms has or even more ... it's really sad that cppcms almost died ... On Fri, Oct 28, 2016 at 5:21 PM, <ano...@op...> wrote: > Just what I want add - I am very motivated to improve CppCMS and port it > to C++11, and all what needed - new branch, where I can send patches. > > ------------------------------------------------------------ > ------------------ > The Command Line: Reinvented for Modern Developers > Did the resurgence of CLI tooling catch you by surprise? > Reconnect with the command line and become more productive. > Learn the new .NET and ASP.NET CLI. Get your free copy! > http://sdm.link/telerik > _______________________________________________ > Cppcms-users mailing list > Cpp...@li... > https://lists.sourceforge.net/lists/listinfo/cppcms-users > |
|
From: <ano...@op...> - 2016-10-28 14:22:04
|
Just what I want add - I am very motivated to improve CppCMS and port it to C++11, and all what needed - new branch, where I can send patches. |
|
From: CN <cn...@fa...> - 2016-10-28 14:07:07
|
On Fri, Oct 28, 2016, at 06:10 AM, ano...@op... wrote: > Hi Artyom, > > I read your position about migration to C++11 > > https://sourceforge.net/p/cppcms/mailman/message/35292236/ > > BTW, I have proposal to preparing release CppCMS 2.0 with breaking API. > > Rationale: > > 1. We can migrate to C++11 (or C++14). > > 2. We can drop some booster classes, and using pure STL and new language > features. For example, migrate to std::thread, smart-pointers, etc. > > 3. We can clean-up code, make it more robust, cleaner and faster (rvo, > constexpr, lambdas, syntax sugar). > > 4. We can completely drop old and deprecated stuff. > > 5. We can continue support 1.x for those who need C++0x. > > [*we - you and community] > > > Of course, it will little bit harder because of needing to backport > bugfixes and some features into 1.x. > > If you don't want do it, then please answer - how many years you want to > wait before CppCMS will migrate to C++11? > > Regards. > I prefer to not placing too much weight on Artyom - the sole developer of this marvelous project. I asked myself from time to time this question: "How could it be possible that a single man can complete a full functioning framework in large scale of so high quality like CppCMS?" However, I am not saying that I am absolutely comfortable with the quality of CppCMS documentation. In fact, I sometimes spent hours reading the documentation struggling to understand the usage of only one class method, but in the end I had to guess and test it in order to get an idea of its functionality. Therefore, if I can vote, I will place improving the quality of documentation as the first entry in the to-do list. My this comment then leads to one question: Why didn't I improve the documentation for the benefit of other users in this community after I finally understood the meanings of that part of documentation? Here are my "reasons": - English is not my native speaking language. - I do not have sufficient confidence on my technical knowledge on CppCMS to *correctly* improve its documentation. I can not rule out the possibility of misunderstanding CppCMS components. - I have been too busy to feed back to this community. My these reasons, "excuses" in another word, in fact are where the problems really lie - If every CppCMS community member has his/her own reasons for which he/she is unable to contribute to this project, it is not good. Lacking community contributor, the pace of project improvement being slow is one obvious outcome. Worse is the biggest threat of this project's future - an one-man project. I do not know why we have only one developer. I can only guess the possible cause - the adoption of this project has not been large due to the following possible "trends": 1. Many web programmers flood to OS tied applications for portable devices. 2. Many web programmers embrace NODE JS. 3. The learning curve of C++ is too steep for many web programmers. I have no idea what I can do about these "trends". Perhaps the best we community members can do at this moment is making your own CppCMS applications great - so great that they make your businesses so successful that your organizations grow and expand and therefore begin to hire top end programmers to work exclusively on CppCMS framework. When this comes true, it means that the CppCMS adoption grows. Once the adoption grows, real good programmers will start to contribute to CppCMS project. Most important of all, do not let this bright future happen too late. Please pardon me if you do not agree with my view points! After all, this is the 2 cents all I have at this moment. Best Regards, CN -- http://www.fastmail.com - Faster than the air-speed velocity of an unladen european swallow |
|
From: <ano...@op...> - 2016-10-27 22:10:34
|
Hi Artyom, I read your position about migration to C++11 https://sourceforge.net/p/cppcms/mailman/message/35292236/ BTW, I have proposal to preparing release CppCMS 2.0 with breaking API. Rationale: 1. We can migrate to C++11 (or C++14). 2. We can drop some booster classes, and using pure STL and new language features. For example, migrate to std::thread, smart-pointers, etc. 3. We can clean-up code, make it more robust, cleaner and faster (rvo, constexpr, lambdas, syntax sugar). 4. We can completely drop old and deprecated stuff. 5. We can continue support 1.x for those who need C++0x. [*we - you and community] Of course, it will little bit harder because of needing to backport bugfixes and some features into 1.x. If you don't want do it, then please answer - how many years you want to wait before CppCMS will migrate to C++11? Regards. |
|
From: Artyom B. <art...@gm...> - 2016-10-09 05:19:06
|
> Many thanks for your kind help!
>
> Now I remember I configured the proxy (my sole client) to always send only
> https requests to my host.
>
> Is there any chance that the following scenario could possibly happen?
>
> Lighttpd somehow did not provide services - maybe temporarily brought down
> by myself. As a result, all requests from proxy rushed to CppCMS process.
> Because the parameter "service.api" is configured as "fastcgi", CppCMS
> process tried to ignore all http requests. However, because the http server
> embed in CppCMS did not recognize https, it therefore did not completely
> close the https connections with proxy.
See, quote from the configuration file"
"service":{
"api":"fastcgi",
"socket":"/tmp/home.sock",
"disable_global_exit_handling":true
}
It means that CppCMS:
1. Brings up ONLY FastCGI module and HTTP is turned off
2. It works over UNIX domain sockets, not TCP/IP (that https uses),
basically CppCMS does not open or close ANY tcp/ip connection so it
must be your's :-)
Artyom Beilis
|
|
From: CN <cn...@fa...> - 2016-10-08 17:16:51
|
Many thanks for your kind help! Now I remember I configured the proxy (my sole client) to always send only https requests to my host. Is there any chance that the following scenario could possibly happen? Lighttpd somehow did not provide services - maybe temporarily brought down by myself. As a result, all requests from proxy rushed to CppCMS process. Because the parameter "service.api" is configured as "fastcgi", CppCMS process tried to ignore all http requests. However, because the http server embed in CppCMS did not recognize https, it therefore did not completely close the https connections with proxy. Best Regards, CN On Sat, Oct 8, 2016, at 01:12 AM, Artyom Beilis wrote: > Hi, > From what I see > 1. The open connections you see aren't ones created by the cppcms > framework itself. The communication with the server actually uses > Unix domain sockets rather than tcp/ip. > 2. HTTP. Timeout affects only HTTP connections, not fastcgi as the > sever is already handles them. > 3. Indeed check your client. -- http://www.fastmail.com - Does exactly what it says on the tin |
|
From: Artyom B. <art...@gm...> - 2016-10-07 17:12:30
|
Hi, >From what I see 1. The open connections you see aren't ones created by the cppcms framework itself. The communication with the server actually uses Unix domain sockets rather than tcp/ip. 2. HTTP. Timeout affects only HTTP connections, not fastcgi as the sever is already handles them. 3. Indeed check your client. Artyom Beilis בתאריך 7 באוק׳ 2016 08:59, "CN" <cn...@fa...> כתב: > > On Wed, Oct 5, 2016, at 04:39 PM, Artyom Beilis wrote: > > Another reason I can see is that your async app does not handle > > timeouts/disconnects properly and keeps connection alive. > > This program does not mount any asynchronous application. > > But your enlightenment does lead me to suspect my synchronous > applications. Some methods in these synchronous applications do not > return until they acquire outside resources like Google re-captcha. > Probably this the cause of the large number of CLOSE_WAIT. > > On the other hand, even the delayed (or never) responses from outside > resources might be the cause to blame, I still have no idea why > http.timeout mechanism did not proactively close the long idle open > connections. > > Best Regards, > CN > > -- > http://www.fastmail.com - Access your email from home and the web > > > ------------------------------------------------------------------------------ > Check out the vibrant tech community on one of the world's most > engaging tech sites, SlashDot.org! http://sdm.link/slashdot > _______________________________________________ > Cppcms-users mailing list > Cpp...@li... > https://lists.sourceforge.net/lists/listinfo/cppcms-users |
|
From: CN <cn...@fa...> - 2016-10-07 05:57:42
|
On Wed, Oct 5, 2016, at 04:39 PM, Artyom Beilis wrote: > Another reason I can see is that your async app does not handle > timeouts/disconnects properly and keeps connection alive. This program does not mount any asynchronous application. But your enlightenment does lead me to suspect my synchronous applications. Some methods in these synchronous applications do not return until they acquire outside resources like Google re-captcha. Probably this the cause of the large number of CLOSE_WAIT. On the other hand, even the delayed (or never) responses from outside resources might be the cause to blame, I still have no idea why http.timeout mechanism did not proactively close the long idle open connections. Best Regards, CN -- http://www.fastmail.com - Access your email from home and the web |
|
From: CN <cn...@fa...> - 2016-10-07 05:40:18
|
On Wed, Oct 5, 2016, at 04:37 PM, Artyom Beilis wrote:
> It should.
>
> Several questions:
>
> 1. Who is the client - are these users or PROXY server?
CppCMS program runs in my host, which is behind CloudFlare's PROXY
network:
- Clients (browsers?) send http or https requests to my host.
- CloudFlare sends these requests to my host on behalf of these clients.
- These requests arrive at my host. Httpd converts http to https.
- CppCMS program responds to CloudFlare.
- CloudFlare responds results to clients using https.
> 2. What is the protocol? it appears as https in the list - is it
> HTTP/FastCGI or SGI
HTTPS/FastCGI
> 3. Post your my-confg.js - do you use http timeouts, what are they?
{
"service":{
"api":"fastcgi",
"socket":"/tmp/home.sock",
"disable_global_exit_handling":true
},
"session":{
"location":"client",
"expire":"browser",
"timeout":72000,
"cookies":{
"prefix":"home"
},
"client":{
"cbc":"aes",
"cbc_key":"my cbc key",
"hmac":"sha1",
"hmac_key":"my hmac key"
}
},
"daemon":{
"enable":false,
"lock":"/var/run/home.pid",
"user":"user2",
"group":"group2"
},
"http":{
"script_names":["/home"]
},
"cache":{
"backend":"thread_shared",
"limit":100
},
"security":{
"csrf":{"enable":true}
},
"logging":{
"level":"debug",
"stderr":false,
"syslog":{
"enable":true,
"id":"home"
}
}
}
Parameter "http.timeout" is not specified in configuration file. I
assume it defaults to 30 seconds according to
http://cppcms.com/wikipp/en/page/cppcms_1x_config#http.timeout
Last time when there was huge number of CLOSE_WAIT, those open
connections appeared to remain open regardless of the 30-second timeout.
They seemed to be not automatically closed by the time out mechanism.
> 4. What version of CppCMS do you use
version 1.2
> 5. Can you connect with strace to the server and see how typical
> connection behave - i.e. see what system calls related to a socket
> executed.
After I re-started CppCMS, I can no longer reproduce this problem -
"lsof" now prints only one CLOSE_WAIT entry. I know this is not
professional, but I will postpone trying strace until when the number of
CLOSE_WAIT surges again.
Best Regards,
CN
--
http://www.fastmail.com - Does exactly what it says on the tin
|
|
From: Artyom B. <art...@gm...> - 2016-10-05 08:39:45
|
Another reason I can see is that your async app does not handle timeouts/disconnects properly and keeps connection alive. Artyom On Wed, Oct 5, 2016 at 11:37 AM, Artyom Beilis <art...@gm...> wrote: > It should. > > Several questions: > > 1. Who is the client - are these users or PROXY server? > 2. What is the protocol? it appears as https in the list - is it > HTTP/FastCGI or SGI > 3. Post your my-confg.js - do you use http timeouts, what are they? > 4. What version of CppCMS do you use > 5. Can you connect with strace to the server and see how typical > connection behave - i.e. see what system calls related to a socket > executed. > > Artyom > >> >> Process# 27567 comes from the running CppCMS program: >> ./cppcms-app -c my-config.js >> >> Is it possible that CppCMS does not cleanly close files? >> >> Best Regards, >> CN >> >> -- >> http://www.fastmail.com - Access your email from home and the web >> >> >> ------------------------------------------------------------------------------ >> _______________________________________________ >> Cppcms-users mailing list >> Cpp...@li... >> https://lists.sourceforge.net/lists/listinfo/cppcms-users |
|
From: Artyom B. <art...@gm...> - 2016-10-05 08:37:17
|
It should. Several questions: 1. Who is the client - are these users or PROXY server? 2. What is the protocol? it appears as https in the list - is it HTTP/FastCGI or SGI 3. Post your my-confg.js - do you use http timeouts, what are they? 4. What version of CppCMS do you use 5. Can you connect with strace to the server and see how typical connection behave - i.e. see what system calls related to a socket executed. Artyom > > Process# 27567 comes from the running CppCMS program: > ./cppcms-app -c my-config.js > > Is it possible that CppCMS does not cleanly close files? > > Best Regards, > CN > > -- > http://www.fastmail.com - Access your email from home and the web > > > ------------------------------------------------------------------------------ > _______________________________________________ > Cppcms-users mailing list > Cpp...@li... > https://lists.sourceforge.net/lists/listinfo/cppcms-users |
|
From: CN <cn...@fa...> - 2016-09-28 06:47:33
|
/var/log/syslog: Sep 28 06:01:51 host1 cppcms-app: cppcms: Caught exception [Failed to read /dev/urandom]#0120x4302ad: booster::runtime_error::runtime_error(std::string const&) + 0x5d in /usr/local/bin/cppcms-app#0120x7fa3e52ea5e4: cppcms::urandom_device::generate(void*, unsigned int) + 0xa4 in /usr/local/lib/libcppcms.so.1#0120x7fa3e52ed961: cppcms::session_interface::generate_csrf_token() + 0x31 in /usr/local/lib/libcppcms.so.1#0120x7fa3e52f2e37: cppcms::session_interface::save() + 0x257 in /usr/local/lib/libcppcms.so.1#0120x7fa3e52694a3: cppcms::http::response::write_http_headers(std::ostream&) + 0x23 in /usr/local/lib/libcppcms.so.1#0120x7fa3e526b1f2: cppcms::http::response::out() + 0x542 in /usr/local/lib/libcppcms.so.1#0120x7fa3e527d024: cppcms::application::render(std::string, cppcms::base_content&) + 0x44 in /usr/local/lib/libcppcms.so.1#0120x454a40: user::login() + 0x690 in /usr/local/bin/cppcms-app#0120x456ae0: booster::function<void ()>::callable_impl<void, cppcms::url_dispatcher::binder0<user> >::call() + 0x30 in /usr/local/bin/cppcms-app#0120x7fa3e5289f61: ??? + 0xe5289f61 in /usr/local/lib/libcppcms.so.1#0120x7fa3e52828cd: cppcms::url_dispatcher::dispatch(std::string) + 0xdd in /usr/local/lib/libcppcms.so.1#0120x7fa3e524503b: cppcms::application::main(std::string) + 0x2b in /usr/local/lib/libcppcms.so.1#0120x7fa3e528257c: ??? + 0xe528257c in /usr/local/lib/libcppcms.so.1#0120x7fa3e52828cd: cppcms::url_dispatcher::dispatch(std::string) + 0xdd in /usr/local/lib/libcppcms.so.1#0120x42be40: multi_thread::main(std::string) + 0x470 in /usr/local/bin/cppcms-app#0120x7fa3e5275824: cppcms::http::context::dispatch(booster::intrusive_ptr<cppcms::application> const&, std::string const&, bool) + 0xa4 in /usr/local/lib/libcppcms.so.1#0120x7fa3e527678e: cppcms::http::context::dispatch(booster::shared_ptr<cppcms::application_specific_pool> const&, booster::shared_ptr<cppcms::http::context> const&, std::string const&) + 0x7e in /usr/local/lib/libcppcms.so.1#0120x7fa3e52786c3: cppcms::impl::thread_pool::worker() + 0xc3 in /usr/local/lib/libcppcms.so.1#0120x7fa3e566d52a: booster_thread_func + 0x1a in /usr/local/lib/libbooster.so.0#0120x7fa3e36e60a4: ??? + 0xe36e60a4 in /lib/x86_64-linux-gnu/libpthread.so.0#0120x7fa3e39e387d: clone + 0x6d in /lib/x86_64-linux-gnu/libc.so.6#012 (http_context.cpp:336) Sep 28 04:46:15 host1 cppcms-app: cppcms: Caught exception [cppdb::posgresql: failed to connect: could not create socket: Too many open files#012]#012 (http_context.cpp:336) Sep 28 04:46:15 host1 cppcms-app: cppcms: Caught exception [cppdb::posgresql: failed to connect: could not create socket: Too many open files#012]#012 (http_context.cpp:336) Sep 28 04:46:15 host1 cppcms-app: cppcms: Caught exception [cppdb::posgresql: failed to connect: could not create socket: Too many open files#012]#012 (http_context.cpp:336) -------------------- lsof | grep 104.16.51.3 | wc yields 5916 records like these: cppcms-app 27567 27591 root 1013u IPv4 1052388506 0t0 TCP host1:39384->104.16.54.3:https (CLOSE_WAIT) cppcms-app 27567 27591 root 1014u IPv4 1052388558 0t0 TCP host1:39390->104.16.54.3:https (CLOSE_WAIT) cppcms-app 27567 27591 root 1015u IPv4 1052388584 0t0 TCP host1:39394->104.16.54.3:https (CLOSE_WAIT) cppcms-app 27567 27591 root 1016u IPv4 1052388586 0t0 TCP host1:39400->104.16.54.3:https (CLOSE_WAIT) IP 104.16.54.3 is one of CloudFlare's host. -------------------- Process# 27567 comes from the running CppCMS program: ./cppcms-app -c my-config.js Is it possible that CppCMS does not cleanly close files? Best Regards, CN -- http://www.fastmail.com - Access your email from home and the web |
|
From: Artyom B. <art...@gm...> - 2016-08-21 12:40:55
|
Unfortunately not much. I recently got my nightly test partially back on-line after my main test PC died. http://cppcms.com/files/nightly-build-report.html (and as you can see not all works yet) Also it needs deeper upgrade (add windows 7 + MSVC 2015, more up-to-date compilers and OSes as well) I'm very busy at my new (paying) job and with my family. There is not too much to go for the next release but I hadn't complete the test suite for the plugin interface (long waited stuff) and didn't completed the API design review (so all works well together) For any real purpose it is stable version also it isn't called like that. What can I say - any help is more than welcome. Artyom > > Is there any update regarding a new release? > > thanks > -- > Christian Gmeiner, MSc > > https://soundcloud.com/christian-gmeiner > > ------------------------------------------------------------------------------ > _______________________________________________ > Cppcms-users mailing list > Cpp...@li... > https://lists.sourceforge.net/lists/listinfo/cppcms-users |
|
From: Artyom B. <art...@gm...> - 2016-08-21 12:30:57
|
On Fri, Aug 19, 2016 at 12:06 PM, Marcel Hellwig <ke...@co...> wrote: > Hi Artyom, > > I see some points in the code where variadic tempaltes[0] would make sense, e.g. url_dispatcher:assign. > There are some thing I notice in the current trunk version. There is a thing called `generic_handler` and I think the purpose is to replace the old handler* typedefs, right? Why not use something like `template <class... Strings> typedef booster::function<void(Strings...)> generic_handler;` ? It would fit all needings and has always the right number of arguments. You use it in the map function that are declared with v1_2, which is strange, because you drop pre-c++11 compiler, is that intended (I would really welcome it)? > The same thing can be used in the url_mapper:map function, where you did not do such thing until now. > The question is, should this framework go full c++11? Drop c++98 completly (hint: yes! :) ). This is good question. However I'm not going to drop C++ 2003 support: 1. I don't want to break existing code for users that can't upgrade or do not want to (for example rock stable Linux releases that do not ship c++11 enable gcc by default) 2. Specific stuff like variadic templates is actually implementable in terms of C++03 without problems (yes it takes much more code) but this isn't what would require me to switch 3. I do want to make it more C++11 friendly - for example by adding move constructors/assignments There are lots of stuff that can be done by moving to C++11 but if you do it, it also good to switch lots of other stuff to C++11 (like booster::thread, booster::regex) but it can't be done fully. To be honest I prefer to keep it C++03 compatible but with new interfaces supporting C++11 like move constructors. > Also, I would like to participate more, but I don't know really how. I made some patch requests in the past, but I don't like the sourceforge way. > I would like to see CppCMS to become a more community-driven and not a single-person project like now. I really like to use it, but don't get me wrong, > I want to become it even better and more popular. I don't know any framework like this. > You know I really need working hands that can write the code, create unit-tests, improve the platform and more. But until now I got very-very few serious contributions. And when I get them I take them seriously. So it is one man project because I'm a single person who actually does the job. Also I have many people who talk about contributions without actually doing any valuable stuff. I have no problem giving the access to SF repo as long as I see some effort, believe me that after several patches i can provide commit rights to contributors (subject to copyright agreements http://cppcms.com/wikipp/en/page/cppcms_1x_coding_standards#Copyrights) However I hadn't seen too many standing in line ;-) > Keep up the good work and please don't forget to answer my questions ;) > > Regards, > Marcel > Regards, Artyom > > > [0]: https://en.wikipedia.org/wiki/Variadic_template > > ------------------------------------------------------------------------------ > _______________________________________________ > Cppcms-users mailing list > Cpp...@li... > https://lists.sourceforge.net/lists/listinfo/cppcms-users |
|
From: Marcel H. <ke...@co...> - 2016-08-19 09:31:50
|
Hi Artyom, I see some points in the code where variadic tempaltes[0] would make sense, e.g. url_dispatcher:assign. There are some thing I notice in the current trunk version. There is a thing called `generic_handler` and I think the purpose is to replace the old handler* typedefs, right? Why not use something like `template <class... Strings> typedef booster::function<void(Strings...)> generic_handler;` ? It would fit all needings and has always the right number of arguments. You use it in the map function that are declared with v1_2, which is strange, because you drop pre-c++11 compiler, is that intended (I would really welcome it)? The same thing can be used in the url_mapper:map function, where you did not do such thing until now. The question is, should this framework go full c++11? Drop c++98 completly (hint: yes! :) ). Also, I would like to participate more, but I don't know really how. I made some patch requests in the past, but I don't like the sourceforge way. I would like to see CppCMS to become a more community-driven and not a single-person project like now. I really like to use it, but don't get me wrong, I want to become it even better and more popular. I don't know any framework like this. Keep up the good work and please don't forget to answer my questions ;) Regards, Marcel [0]: https://en.wikipedia.org/wiki/Variadic_template |
|
From: Christian G. <chr...@gm...> - 2016-08-16 12:18:15
|
2016-03-23 13:07 GMT+01:00 Artyom Beilis <art...@gm...>: > Great Question! > > In general CppCMS 1.1 beta is almost ready (than 1.2 is just matter of > user feedback) > > If you look at the roadmap: > > http://cppcms.com/wikipp/en/page/cppcms_1x_tasks > > I have few major things: > > 1. Complete plugin testing suite - and its combination with templates > and localization > 2. Write down an example of a new plugin API to make sure I didn't miss a thing > 3. Write a test suite for external session API (and do some updates) > 4. Virtual host support (matter of few hours) > > To do all this I need about 3-4 quiet work days... > I hope I'll be able to be ready before May 1st if nothing goes wrong. > > Additionally currently my nightly tests system is down due to HW > malfunction which also > somewhat creates a bottle neck as there are some stuff to fix and test > > http://cppcms.com/files/nightly-build-report.html > > So I need to buy a new computer to replace my 10 years old PC that > runs the tests. > > Bottom line I hope I'll have some available time soon (during upcoming holidays) > but I can't guarantee. > > In any case current trunk is very stable and besides very new plugin > API and external > sessions API I don't see any issues to work with it directly. > Is there any update regarding a new release? thanks -- Christian Gmeiner, MSc https://soundcloud.com/christian-gmeiner |
|
From: Carlos R. G. <dv...@ps...> - 2016-07-27 21:24:40
|
Hello, I'd like to transmit video data from my C++ application to the web page where it is shown. The data is generated on the fly, as h.264 video encapsulated in an MP4 container. On the web page, an HTML5 video element would receive this data and show the video. Is it feasible to send such a live HTTP stream through CppCMS to an HTML video element? |
|
From: dev 1. <dev...@gm...> - 2016-07-26 19:42:23
|
Hi all, I am developing a distributed scientific application which at a very high level can be viewed as a chat application. One feature of which is where clients can subscribe to a group and all the messages sent to the group should be pushed to all the clients subscribed to the group. For implementing this feature I am planning to use the long polling feature of CppCMS. A high level implementation detail with out any error checking would be : 1. The client through a POST request, requests server to join a group. 2. The server extracts the client IP from the POST request and associates it with the requested group. 3. Whenever a message is sent for a group, the server Pushes the message to all the IP address (clients) associated with the group. The push would be implemented using long waiting asynchronous get requests on the client side, similar to the chat example <http://cppcms.com/wikipp/en/page/cppcms_1x_chat>. Following are my questions regarding the application: 1. Is there a better/ more efficient way to implement the multicast operation? 2. If not, is there a better/ more efficient way to associate clients to groups ? Thanks, Dev |
|
From: Christian G. <chr...@gm...> - 2016-07-14 12:39:36
|
Hi all. I run into a very annoying problem with the internal file server provided by cppcms. I somebody wants to download a file it can happen that no data gets send to the user or after some blocks the download process stops. I am using the Qt5 provided mingw build environment to build cppcms and a dummy application. I have seen if the first byte of the read file is 0xff the stream gets closed. Under linux everything works as expected. greets -- Christian Gmeiner, MSc https://soundcloud.com/christian-gmeiner |
17 messages has been excluded from this view by a project administrator.
