The End of hMailserver as we know it?

Use this forum if you have installed hMailServer and want to ask a question related to a production release of hMailServer. Before posting, please read the troubleshooting guide. A large part of all reported issues are already described in detail here.
Post Reply
User avatar
Dravion
Senior user
Senior user
Posts: 1635
Joined: 2015-09-26 11:50
Location: Germany
Contact:

The End of hMailserver as we know it?

Post by Dravion » 2016-03-18 08:25

I was exploring the repo lately and found a new
Repo on GitHub authored by Martin just a few days ago:

Description:
A very experimental work-in-progress re-implementaiton of hMailServer using .NET.
https://github.com/hmailserver/hmailse ... README.md

Now i understand what he is talking about at the license thread...
The next major version of hMailServer might have a change of license to GPL. But the next version is so different from the current I'm not even sure it will keep its name. 

tohare
Normal user
Normal user
Posts: 181
Joined: 2014-12-07 13:35
Location: Florida!

Re: The End of hMailserver as we know it?

Post by tohare » 2016-03-18 09:03

If the next true version is that different it may be a great idea to rename it. Yes, you need to rebuild a product name on the market but since it is (Still?) a freebie I cannot see it hurting. Of course appending an additional name such as hMailserver Plus (just a name I used for example) you could still keep it's main ID.

Considering the changes in technology plus lessons learned through the years it may be a good choice to rebuild (lessons from my own years of Dev on other projects). Sometimes getting rid of legacy code sections that could have been written better can be a programmer's delight.

The problem with a product overhaul with the same name is you will get a lot of blow back about it being so different from Admins it may be overwhelming. And considering any major rewrite inherently has new issues -- both features and operation -- the feedback may not be as positive as you would like and could affect the product reputation.

Just my 3 cents.
Thanks,
Thomas

User avatar
jimimaseye
Moderator
Moderator
Posts: 8364
Joined: 2011-09-08 17:48

Re: The End of hMailserver as we know it?

Post by jimimaseye » 2016-03-18 09:24

tohare wrote:Of course appending an additional name such as hMailserver Plus (just a name I used for example) you could still keep it's main ID.
It is ESSENTIAL that there is an element of the original product name (the above is a perfect example). Products take a LONG time to get established from new - and no one knows a product just by name alone; a product is *known* by its reputation and the name of the product just serves as an identity for that established product.

As in,
Q: ("What is the name of a really good mail server product that runs on windows that everyone uses?")
A:
  • a, "its hMailserver (and so now where youll be able to find it)" OR
    b, "it used to be hMailserver but now I dont know. Youll have to search for one and test it to see."
If he changes the name completely, and loses any recognition of the "hMailserver" from the name, than the product will need to re-establish and rebuild its reputation again irrespective of how well existing users rate it and how successful/popular it has been in the past.

"Rebranding" is a good toy in the right circumstances in marketing. But often it is more about modifying the *appearance* (logo, colours, font, tag lines, "mission statement" etc) but its important to maintain the recognition of the product by keeping its well-respected roots obvious in the name (.....if the old product WAS "well-respected" of course. If your old product was a pile of pants, then go ahead and completely change the name and all recognition to ditch your bad history and start again and give yourself a chance).
5.7 on test.
SpamassassinForWindows 3.4.0 spamd service
AV: Clamwin + Clamd service + sanesecurity defs : https://www.hmailserver.com/forum/viewtopic.php?f=21&t=26829

User avatar
Dravion
Senior user
Senior user
Posts: 1635
Joined: 2015-09-26 11:50
Location: Germany
Contact:

Re: The End of hMailserver as we know it?

Post by Dravion » 2016-03-18 12:01

Hmm, from a technical perspective its a comletenew server.NET Code is like Java int runs in an Virtual Machine.The otherside is, Windows .NET and OpenSource .NET are not the same and not 100% compatible. So hMailserver.NET will never run on any other OS then Windows.

EOL - End of Life support for native HMS
If the new .NET hMailserver moves on i fear the old version we all know and like will declared as deprecated no longer supported, so you need to migrate to hMailserver .net in the future.

.NET Future itself
At Microsoft it was decided to create a new .NET System called Version .NET Core (Project Roslyn). Its not just a new version, its a new technology with focus on Wundows 10 and the so called Windows Universal Architecture. Of cause, for marketing reasons MS tells everyone Linux will be supported and thats correct but the same old song plays once again: Its not 100% compatible and what runs on Windows 10 only doesnt run anywhere else.

Kelden
Normal user
Normal user
Posts: 43
Joined: 2011-05-02 13:58

Re: The End of hMailserver as we know it?

Post by Kelden » 2016-03-18 13:13

So hMailserver.NET will never run on any other OS then Windows.
Why not? If you stick to .NET core it will run everywhere like with JAVA.

User avatar
SorenR
Senior user
Senior user
Posts: 3414
Joined: 2006-08-21 15:38
Location: Denmark

Re: The End of hMailserver as we know it?

Post by SorenR » 2016-03-18 13:25

Why change the name... I have a program called "Paint"... I have another program called "Paint.NET" - it's sort of the same thing only using different technologies :mrgreen:
SørenR.

“With age comes wisdom, but sometimes age comes alone.”
- Oscar Wilde

User avatar
Dravion
Senior user
Senior user
Posts: 1635
Joined: 2015-09-26 11:50
Location: Germany
Contact:

Re: The End of hMailserver as we know it?

Post by Dravion » 2016-03-18 13:45

Kelden wrote:
So hMailserver.NET will never run on any other OS then Windows.
Why not? If you stick to .NET core it will run everywhere like with JAVA.
Good joke.
hMailserver comes with a GUI Admin tool and for Martin simplicity and a hMailserver version without
GUI Admin is out of discussion for Martin. The Problem with .NET Core on Linux is, it doesnt support GUI Programming in any shape or form. Even .NET Core on Linux is not the same as on Windows. Besides this fact, Martin is not targeting .NET -ore at all, he is using the old .NET-Framework and there we facing the same sort of problems. For instance: while .NET-Legac GUI-Programs on Windows are not a big deal (Winforms, WPF, XAML) its a complete diffrent story on Linux (you have to use native Subsitutes like binding to GTK2 ect). Thats was in the past main reason why no bigger Windows .NET Projects has benever sucessfully portet to .NET Mono. Even on Windows, big .NET Programs and even Games are very rare. Even Visual Studio who claims to use "some parts of .NET is now going back to native because the overall performance was terrible in compare to Visual Studio 2015, which runs 100 times faster.

Anyhow:
I dont expect any will or motivation from Martin to support .NET/.NET-Core on any other OS then Windows so this debate (hMailServer.NET on Linux) doesnt make any sense at all.

User avatar
jimimaseye
Moderator
Moderator
Posts: 8364
Joined: 2011-09-08 17:48

Re: The End of hMailserver as we know it?

Post by jimimaseye » 2016-03-18 13:50

Dravion wrote:So hMailserver.NET will never run on any other OS then Windows.
Maybe Martin has never had intentions of it being compatible with other platforms. There are plenty of Linux offerings already.
5.7 on test.
SpamassassinForWindows 3.4.0 spamd service
AV: Clamwin + Clamd service + sanesecurity defs : https://www.hmailserver.com/forum/viewtopic.php?f=21&t=26829

User avatar
Dravion
Senior user
Senior user
Posts: 1635
Joined: 2015-09-26 11:50
Location: Germany
Contact:

Re: The End of hMailserver as we know it?

Post by Dravion » 2016-03-18 14:28

Maybe Martin has never had intentions of it being compatible with other platforms. There are plenty of Linux offerings already.
Youre right but all the so called offerings sharing the same problems.

1) No SMTP/POP3/IMAP integrated solution.
2) Text-Configfiles (in most cases no GUI at all)
3) Heavy learning curve to master postfix and co
4) Not suitable for the average user.

It would be great to have at least one "simple & userfriendly" opensource mailserver out there, running on windows, mac, linux/unix.

For technical experts there are plenty of options but not for the rest of us.

User avatar
jimimaseye
Moderator
Moderator
Posts: 8364
Joined: 2011-09-08 17:48

Re: The End of hMailserver as we know it?

Post by jimimaseye » 2016-03-18 14:35

Dravion wrote: For technical experts there are plenty of options but not for the rest of us.
Love how you dont classify yourself as an 'expert', Dravion. Very modest of you.
5.7 on test.
SpamassassinForWindows 3.4.0 spamd service
AV: Clamwin + Clamd service + sanesecurity defs : https://www.hmailserver.com/forum/viewtopic.php?f=21&t=26829

Kelden
Normal user
Normal user
Posts: 43
Joined: 2011-05-02 13:58

Re: The End of hMailserver as we know it?

Post by Kelden » 2016-03-18 14:50

@Dravion

Who cares about GUI and hMS core as a single project? You could split and and stick with the engine to .NET core and do whatever you want with the GUI. Not a big deal. I think you have never done any .NET development.
And Visual Studio is a WPF application and there were a lot of improvements to WPF because they startet using it in VS.

User avatar
Dravion
Senior user
Senior user
Posts: 1635
Joined: 2015-09-26 11:50
Location: Germany
Contact:

Re: The End of hMailserver as we know it?

Post by Dravion » 2016-03-18 15:04

Kelden wrote:@Dravion

Who cares about GUI and hMS core as a single project? You could split and and stick with the engine to .NET core.
You should try to read first before you post.

GUI was and is essential for hMailServer. Our GUI Admin is cimplete in .NET for years. Martin (the Maindeveloper) has decided that the GUI is a core element of hMailserver.You will never see an official hMailserver releasw without it. If you dont like it use postfix,sendmail or exim.

Kelden
Normal user
Normal user
Posts: 43
Joined: 2011-05-02 13:58

Re: The End of hMailserver as we know it?

Post by Kelden » 2016-03-18 15:27

@Dravion.

You still don't understand. GUI is important. But currently there is also the hMS service and the GUI.
The service needs no GUI so it's easy to stick to .NET core.

User avatar
jimimaseye
Moderator
Moderator
Posts: 8364
Joined: 2011-09-08 17:48

Re: The End of hMailserver as we know it?

Post by jimimaseye » 2016-03-18 15:45

Kelden wrote:The service needs no GUI so it's easy to stick to .NET core.
How do you configure accounts? View the Server Status? View the current delivery queue? Monitor the logging in real time? View the current IP RANGE/bans?

No point having a service that is ready for connections that hasnt been configured to do so. (And please do go down the 'do it like Linux and use text config files' route; the one thing that makes HMS stand out from other solutions is the GUI (accessibility and simplicity)....otherwise as a service (wirhout a GUI) it would just be *another mailserver* sending and receiving emails. And there are lots of them so why would we bother with Hmailserver? (Think before answering with something like "use the PHP admin console" or 'web interface' : that would be a GUI then!)
5.7 on test.
SpamassassinForWindows 3.4.0 spamd service
AV: Clamwin + Clamd service + sanesecurity defs : https://www.hmailserver.com/forum/viewtopic.php?f=21&t=26829

Kelden
Normal user
Normal user
Posts: 43
Joined: 2011-05-02 13:58

Re: The End of hMailserver as we know it?

Post by Kelden » 2016-03-18 16:01

How do you configure accounts? View the Server Status? View the current delivery queue? Monitor the logging in real time? View the current IP RANGE/bans?
How you do it now? Write it to the database with the GUI?
Or connect the GUI to the service and tell the service the settings you want to change.
No change how it works now.

User avatar
mattg
Moderator
Moderator
Posts: 20632
Joined: 2007-06-14 05:12
Location: 'The Outback' Australia

Re: The End of hMailserver as we know it?

Post by mattg » 2016-03-18 16:07

Kelden wrote:How you do it now? Write it to the database with the GUI?
no
Kelden wrote:Or connect the GUI to the service and tell the service the settings you want to change.
This one
Just 'cause I link to a page and say little else doesn't mean I am not being nice.
https://www.hmailserver.com/documentation

User avatar
SorenR
Senior user
Senior user
Posts: 3414
Joined: 2006-08-21 15:38
Location: Denmark

Re: The End of hMailserver as we know it?

Post by SorenR » 2016-03-18 16:17

The GUI is just a shell connecting to the API to help us remember what we have forgotten... What ever programming language the GUI is made with should be irellevant as long as it support the API...

If you use e.g. a REST XML http based (or what the h*ll you call it) you are free to create a GUI to run on VMS, AS/400, Linux, Windows, iOS, Android, Commodore 64 - whatever... As simple as that.
SørenR.

“With age comes wisdom, but sometimes age comes alone.”
- Oscar Wilde

User avatar
jimimaseye
Moderator
Moderator
Posts: 8364
Joined: 2011-09-08 17:48

Re: The End of hMailserver as we know it?

Post by jimimaseye » 2016-03-18 16:32

SorenR wrote:The GUI is just a shell connecting to the API to help us remember what we have forgotten... What ever programming language the GUI is made with should be irellevant as long as it support the API...

If you use e.g. a REST XML http based (or what the h*ll you call it) you are free to create a GUI to run on VMS, AS/400, Linux, Windows, iOS, Android, Commodore 64 - whatever... As simple as that.
....indeed. But it is still essential to HMS to have one (bringing it back to comments further above).
5.7 on test.
SpamassassinForWindows 3.4.0 spamd service
AV: Clamwin + Clamd service + sanesecurity defs : https://www.hmailserver.com/forum/viewtopic.php?f=21&t=26829

Kelden
Normal user
Normal user
Posts: 43
Joined: 2011-05-02 13:58

Re: The End of hMailserver as we know it?

Post by Kelden » 2016-03-18 16:34

For me it's also important. I just want to say that hMS "could" run on Linux when the service is done with .NET Core.
For me there is no discussion that we need a GUI.

User avatar
SorenR
Senior user
Senior user
Posts: 3414
Joined: 2006-08-21 15:38
Location: Denmark

Re: The End of hMailserver as we know it?

Post by SorenR » 2016-03-18 16:49

Kelden wrote:For me it's also important. I just want to say that hMS "could" run on Linux when the service is done with .NET Core.
For me there is no discussion that we need a GUI.
Why not compile the OS into hMailServer so it runs directly on a Hyper-V core... Job done...
SørenR.

“With age comes wisdom, but sometimes age comes alone.”
- Oscar Wilde

Kelden
Normal user
Normal user
Posts: 43
Joined: 2011-05-02 13:58

Re: The End of hMailserver as we know it?

Post by Kelden » 2016-03-18 17:17

Why not compile the OS into hMailServer so it runs directly on a Hyper-V core... Job done...
Here is your award for the most stupid answer today.

User avatar
SorenR
Senior user
Senior user
Posts: 3414
Joined: 2006-08-21 15:38
Location: Denmark

Re: The End of hMailserver as we know it?

Post by SorenR » 2016-03-18 17:27

Kelden wrote:
Why not compile the OS into hMailServer so it runs directly on a Hyper-V core... Job done...
Here is your award for the most stupid answer today.
What did I win ??
SørenR.

“With age comes wisdom, but sometimes age comes alone.”
- Oscar Wilde

User avatar
jimimaseye
Moderator
Moderator
Posts: 8364
Joined: 2011-09-08 17:48

Re: The End of hMailserver as we know it?

Post by jimimaseye » 2016-03-18 17:32

Image

:mrgreen:
5.7 on test.
SpamassassinForWindows 3.4.0 spamd service
AV: Clamwin + Clamd service + sanesecurity defs : https://www.hmailserver.com/forum/viewtopic.php?f=21&t=26829

User avatar
Dravion
Senior user
Senior user
Posts: 1635
Joined: 2015-09-26 11:50
Location: Germany
Contact:

Re: The End of hMailserver as we know it?

Post by Dravion » 2016-03-18 18:37

Hmm, maybe there is an way for targetting .NET-Core but with the option to stay independent from it if needed.

.NET-Core
Its the new Kid on the block and will exists in parallel and will replace the old .NET in the future. Old .NET is a VM based system (like Java) and our GUI Client is using it. However: the new .NET-version is not an VM-System, its more like an crossplatform runtime system. It can be programmed with various languages like C/C++, C# and even Visual Basic but the the diffrence is, its not managed code/bytecode, its binary code.

IMHO:I t should be possible compiling one version of hmailserver.exe as .NET-Core version while a second build can be compiled as normal, native one which doesnt need .NET-Core at all. Why? Because on Windows .NET-Core is the future and in my view it makes no sense writing or rewriting a program on an end of life framework like old .NET is.

User avatar
mattg
Moderator
Moderator
Posts: 20632
Joined: 2007-06-14 05:12
Location: 'The Outback' Australia

Re: The End of hMailserver as we know it?

Post by mattg » 2016-03-19 00:14

Kelden wrote:
Why not compile the OS into hMailServer so it runs directly on a Hyper-V core... Job done...
Here is your award for the most stupid answer today.
I'm intrigued by why you think this is a stupid answer.

hMailserver in it's current form GUI and all, will run on Hyper-V Core.
I've had it running on Server Hyper-V Core 2012, 2012R2 and 2014

Needed to download dot net framework for the GUI to work, but could have as easily install Apache and PHP, and used PHPWebAdmin to manage.
SorenR wrote:The GUI is just a shell connecting to the API to help us remember what we have forgotten... What ever programming language the GUI is made with should be irellevant as long as it support the API...

If you use e.g. a REST XML http based (or what the h*ll you call it) you are free to create a GUI to run on VMS, AS/400, Linux, Windows, iOS, Android, Commodore 64 - whatever... As simple as that.
Martin has already stated that he is looking to change the gui to something platform independent

viewtopic.php?f=10&t=26811&p=165179&hil ... ui#p165179
Just 'cause I link to a page and say little else doesn't mean I am not being nice.
https://www.hmailserver.com/documentation

tohare
Normal user
Normal user
Posts: 181
Joined: 2014-12-07 13:35
Location: Florida!

Re: The End of hMailserver as we know it?

Post by tohare » 2016-03-19 08:21

Glad it ain't me as I have wanted to look for something other than .Net and a Command Line Interface (CLI). I am neither a fan of MS or .Net for a number of reasons and don't want to get into it. I love Linux but a CLI is just plain clunky (IMO). A CLI just seems like the DOS days and really don't want to move backward. A picture is worth a 1k words said the Chinese genius and GUI makes it much faster to navigate and get stuff done than using a CLI. And programming w/o intellisense and tool tips, etc, (GUI) would "slow" development.

Yes I would love to see a cross platform version just because it is much easier to learn 1 package than 2. And a CLI on one and GUI on other would kill the reason just specified previously.

I have been chasing this monster around for a while so when you guys get the perfect answer ignite some flares so I can see the single solution. Just remember MS is now porting a version of SQL Server to Linux. To me, that means the Linux server market is kicking MS servers in the butt very hard -- so expect more Linux stuff.

But I like HMS as it is decent, a little clunky to install, but it gets the job done in less than Enterprise environments.

In the mean time leave me out of the flame wars -- PLEASE!!!! :mrgreen:
Thanks,
Thomas

User avatar
Dravion
Senior user
Senior user
Posts: 1635
Joined: 2015-09-26 11:50
Location: Germany
Contact:

Re: The End of hMailserver as we know it?

Post by Dravion » 2016-03-19 09:48

Its not only the personal preferences its the whole thing that martin seems to give a shit about the community and decides something important just on his own. For me, i was working on a port to Linux and Mac but iam not willing to do it anymore if the entire thing is now switched over to legacy windows .NET.

Of cause i could fork and do my own thing but it would be the same thing as two diffrent projects, two diffrent communities ect. Sorey but iam pissed.

User avatar
jimimaseye
Moderator
Moderator
Posts: 8364
Joined: 2011-09-08 17:48

Re: The End of hMailserver as we know it?

Post by jimimaseye » 2016-03-19 12:11

Dravion wrote:Its not only the personal preferences its the whole thing that martin seems to give a shit about the community and decides something important just on his own
I just wonder where the notion of HMailserver is "the people's software" comes from. Im relatively new to HMS (4 years), and Im not very good with politics. But from what I see, read and understand from the past, this has always been Martin's concept, Martins project and Martin's decision on what is included and how it is to be put together.

I suspect the involvement of it being 'open' to you on Github is (understandably) misinterpreted. The benefit is that you can take it and do what you want for your own benefits (download the source and do what you want), and maybe you could even offer amendments, but ultimately Martin continues do his thing and do it in the direction he chooses as it is his project. As I see it, this is no different from other 'open source' software such as, for example, the Mozilla offerings (FF and TB etc) - sure you can take the source, and offer objections and suggestions, but ultimately it's the key development team (of Mozilla) that decides the direction of the product. It really isnt a 'public contribution' project and doesnt look like it ever has been (save for feedback and suggestion).

Im sure there are those that may try to counter my interpretation with an official stance of HMS similar to "open source, therefore contribution allowed etc". Im not talking about the official stance (if indeed it is different from my interpretation), Im talking about the reality. 'Developer Bill' had an input in to HMS but ultimately it seems Martin always had the final decision about what and how to do things (a clash of personalities and methods ensued and Bill 'left', it seems). And it's not the first time I have had a discussion about these things and I am certainly not the only one seeing it in this way.

It is all about "perception". My perception is that you do not have a control about the direction of the product (and that most suggestions/votes will be ignored). Bills perception was that he was allowed to contribute and modify and have his code included in to product. Your perception is you should have some say on the platform and the direction of its future.

Perhaps if Martin kept the product a 'closed source' project that you had no access to then you wouldnt be so pissed off. You would download it, try it, like it/accept it OR dislike it/try another product (and then come back to HMS again :wink: ), and not feel so involved in it's future (and save you from this disappointment you feel).

I guess we just have to accept the product for what it is. Step forward, do your feedbacks and suggestions on Github, and then just step back again.
5.7 on test.
SpamassassinForWindows 3.4.0 spamd service
AV: Clamwin + Clamd service + sanesecurity defs : https://www.hmailserver.com/forum/viewtopic.php?f=21&t=26829

User avatar
SorenR
Senior user
Senior user
Posts: 3414
Joined: 2006-08-21 15:38
Location: Denmark

Re: The End of hMailserver as we know it?

Post by SorenR » 2016-03-19 13:30

jimimaseye wrote:
Dravion wrote:Its not only the personal preferences its the whole thing that martin seems to give a shit about the community and decides something important just on his own
I just wonder where the notion of HMailserver is "the people's software" comes from. Im relatively new to HMS (4 years), and Im not very good with politics. But from what I see, read and understand from the past, this has always been Martin's concept, Martins project and Martin's decision on what is included and how it is to be put together.

bla bla bla...
Jimmy, I think you and I are on the same page here and what is more important - WE are NOT "children of the Internet"...

I see it in politics and I see it in parenting...

Hell no, I was in my late twenties when the Internet was invented and it was still legal to spank naughty children in school :mrgreen:
SørenR.

“With age comes wisdom, but sometimes age comes alone.”
- Oscar Wilde

User avatar
mattg
Moderator
Moderator
Posts: 20632
Joined: 2007-06-14 05:12
Location: 'The Outback' Australia

Re: The End of hMailserver as we know it?

Post by mattg » 2016-03-19 14:46

SorenR wrote:Jimmy, I think you and I are on the same page here
ditto
Just 'cause I link to a page and say little else doesn't mean I am not being nice.
https://www.hmailserver.com/documentation

User avatar
Dravion
Senior user
Senior user
Posts: 1635
Joined: 2015-09-26 11:50
Location: Germany
Contact:

Re: The End of hMailserver as we know it?

Post by Dravion » 2016-03-19 15:34

mattg wrote:
SorenR wrote:Jimmy, I think you and I are on the same page here
ditto
iam not.

My perception of OpenSource is indeed a complete diffrent perspective.
Enhance, patch, fix for the better of all users and the development process is what keeps me going.
But i guess (as jimi descibed) Martins thinks diffrent about it. Lessons learned. I think iam better of with a new approach,a community drivven OpenSource Mailserver project which supports Windows, Linux, Unix, Mac and Userfriendly from the start, where the community, not the maindeveloper has the last say.
I was a little bit pissed, but now its ok for me. hMailserver is a great product and maybe the new .NET version brings some fresh air into it :wink:

User avatar
martin
Developer
Developer
Posts: 6834
Joined: 2003-11-21 01:09
Location: Sweden
Contact:

Re: The End of hMailserver as we know it?

Post by martin » 2016-03-19 21:17

Interesting and slightly confusing read, so I'm not sure where to start. :) I mean the GitHub repo description even says it's an experiment, so it seems like you are reading a bit too much into it. I assume I've been too unclear. The intention is not that the code in the hmailserver-net repo will replace hMailServer. I won't be releasing a hMailServer 6 which is a complete rewrite from scratch on another runtime platform. I have no intention to just drop the old version and say this no longer exists and I won't patch any bug you find (I assume that's to some degree what you mean by "supported").
Besides this fact, Martin is not targeting .NET -ore at all, he is using the old .NET-Framework and there we facing the same sort of problems.
Yes. The reason for that is that the tooling support is not mature yet for my liking. If it were, I would have targeted .NET Core. I mean, .NET Core isn't even available as a real release, just Release Candidates as of today, right? And I don't have enough spare time to fiddle around with MS-stuff which isn't actually released.
martin seems to give a shit about the community and decides something important just on his own.
If I feel like setting up a new github repo where I implement some part of hMailServer in a completely new way on my spare time and not sponsored by any users, then you're right I don't give a shit what other people think about that. What I spend my spare time on is not a community decision. hMailServer has been open source for almost 15 years (with a few versions closed for reasons related to my own stupidity). Except for 3rd party libraries such as Boost and OpenSSL, >97% of what is in hMailServer today has been written by me. The number of other people contributing to the hMailServer source code (I mean actual C++/C# code) has been less than ~0.5 per year. To be honest - I only remember 4 people actually contributing anything significant so far. With this in mind, maybe you don't find it so strange that I don't ask other people for permission before doing something. Of course the development of hMailServer isn't just about the code but also about the community and the great people around here who spend their time helping other users, but these have rarely contacted me about code-level opinions.

The reason I put up the other github repo was because I like to try out new things and experiment. There are actually some major benefits which could be made by implementing a mail server in a very different way than hMailServer is implemented. When I put up the new repo, I was not sure what to call it (I thought about "MailServerKit", "MailServerToolbox" etc, but I didn't come up with a good name so I just appended -net to the end, which was probably a bad decision in retrospect.
Martin Knafve
martin@hmailserver.com
https://twitter.com/knafve

User avatar
jimimaseye
Moderator
Moderator
Posts: 8364
Joined: 2011-09-08 17:48

Re: The End of hMailserver as we know it?

Post by jimimaseye » 2016-03-19 21:32

Thanks Martin, hopefully that has cleared things up for people.

p.s I am really genuinely sorry I am not able to do 'coding' - I would LOVE to help the project (which would have made the contributor count 4½ :wink: ). The benefits would be that a, I like the idea of being 'useful' and b, I really like the idea of the hMailserver evolution (releases with bug fixes and new features) being a little quicker than it is. (Understandably, being only 1 man doing it all, it can only evolve at the pace you go at).

Maybe @Dravion, who seems a very knowledgeable man, is 'coding confident' and can offer his contributions.
5.7 on test.
SpamassassinForWindows 3.4.0 spamd service
AV: Clamwin + Clamd service + sanesecurity defs : https://www.hmailserver.com/forum/viewtopic.php?f=21&t=26829

User avatar
martin
Developer
Developer
Posts: 6834
Joined: 2003-11-21 01:09
Location: Sweden
Contact:

Re: The End of hMailserver as we know it?

Post by martin » 2016-03-19 21:36

Well, the code is of course important but hMailServer would not exist today if it weren't for other people who endlessly have the patience to help other users. People helping out in the forum is just as valuable to the community as people submitting patches. But I agree, the rate of releases is not so high currently. Last few weeks I've tried to spend some time every day on hMailServer and hopefully that will have some impact (which happens to be no a laptop on the bus to work, because this is pretty much the only free time I have during the day).
Martin Knafve
martin@hmailserver.com
https://twitter.com/knafve

User avatar
jimimaseye
Moderator
Moderator
Posts: 8364
Joined: 2011-09-08 17:48

Re: The End of hMailserver as we know it?

Post by jimimaseye » 2016-03-19 21:42

If you dont mind, just to give us an idea, could you explain your 'set up' (in order we have a better understanding of time constraints). eg, Full time job and HMS is spare time? Coding is your job or coding is hobby? Other coding projects also take up your time or is HMS the only one?
5.7 on test.
SpamassassinForWindows 3.4.0 spamd service
AV: Clamwin + Clamd service + sanesecurity defs : https://www.hmailserver.com/forum/viewtopic.php?f=21&t=26829

User avatar
martin
Developer
Developer
Posts: 6834
Joined: 2003-11-21 01:09
Location: Sweden
Contact:

Re: The End of hMailserver as we know it?

Post by martin » 2016-03-20 21:48

A bit off topic and kind of personal, but I don't mind it. I have a full time job as a software developer working for another company (I do software development, but it's in no way related to hMailServer). So I leave home 6:30 AM and when I get home around 5PM, I take care of my 2 year old son until he falls asleep at around 8:30-9PM so that my girlfriend can study (my son is too young to play on his own for any longer periods of time). So my "spare time" when I can spend time on hMailServer is typically on the bus to and from work, and on evenings after 9PM (but at 9 in the evening my brain starts to shut down).

I don't have time for any other 'big' coding projects, but I sometimes spend time on small stuff, such as https://github.com/martinknafve/Bitifier.RsaEncryption.
Martin Knafve
martin@hmailserver.com
https://twitter.com/knafve

User avatar
jimimaseye
Moderator
Moderator
Posts: 8364
Joined: 2011-09-08 17:48

Re: The End of hMailserver as we know it?

Post by jimimaseye » 2016-03-20 22:16

Well, thank you for the frank picture and clarity (a little bit more personal than I was expecting actually (involving your family life there)). Good to know and yes it really does give us a better idea of how your time is spent on Hmailserver and more of an understanding on the 'delays' between updates (and what to expect from you when contacted personally with crash dumps etc on your martin@ address).
5.7 on test.
SpamassassinForWindows 3.4.0 spamd service
AV: Clamwin + Clamd service + sanesecurity defs : https://www.hmailserver.com/forum/viewtopic.php?f=21&t=26829

User avatar
martin
Developer
Developer
Posts: 6834
Joined: 2003-11-21 01:09
Location: Sweden
Contact:

Re: The End of hMailserver as we know it?

Post by martin » 2016-03-20 22:24

When it comes to crash dumps, if someone reports me with a reproducible crash that will be prioritized. I look at all the crash dumps I receive. The issues with high memory usage isn't really resolved until 5.7.
Martin Knafve
martin@hmailserver.com
https://twitter.com/knafve

User avatar
Dravion
Senior user
Senior user
Posts: 1635
Joined: 2015-09-26 11:50
Location: Germany
Contact:

Re: The End of hMailserver as we know it?

Post by Dravion » 2016-03-20 22:52

Can you explain how you solved it?
Is this a boost:asio thread memory consumption error or a memory leak?

User avatar
martin
Developer
Developer
Posts: 6834
Joined: 2003-11-21 01:09
Location: Sweden
Contact:

Re: The End of hMailserver as we know it?

Post by martin » 2016-03-21 07:45

It's more related to the issue that hMailServer caches things in memory without real regard of how much memory is available. Such as folders, message lists and so on.

In hMailServer 5.7 this has been rewritten so the total memory usage is limited. hMailServer 5.7 is still alpha though, but I've been running it on my own 2 servers for a couple of months without any issues. More stress tests of this area is needed though (the load on my own servers aren't high enough to validate that it works).

Of course, there could be other issues as well causing high memory usage - I've addressed the one I'm aware of.
Martin Knafve
martin@hmailserver.com
https://twitter.com/knafve

User avatar
Dravion
Senior user
Senior user
Posts: 1635
Joined: 2015-09-26 11:50
Location: Germany
Contact:

Re: The End of hMailserver as we know it?

Post by Dravion » 2016-03-21 11:14

Hmm, as far as i know boost::asio is async, so this shouldnt exaust the memory even if you have 10.000 concurrent connects via smtp/pop3/imap
Which pattern do you use for caching in thepast and what pattern/design do you implement or 5.7?

Post Reply