Eggdrop1.6.20rc1 and SSL

[Gargantua]

Planning on installing the new version eggdrop1.6.20rc1 and wanted to know if this version has "built-in" support for SSL connections, or do i have to use a patch like Rootie´s SSL patch?

If so, is the present version of that patch compatible with eggdrop1.6.20rc1? If not, are there a newer version or another one?

[pseudo]

No, eggdrop 1.6.20 doesn't have built-in support for SSL. We will finally add full IPv6 and SSL in the next eggdrop version. The current patch for .19 would need some modifications to apply to .20, but I guess somebody will release an update to it soon.

On the other hand, IPv6 and SSL are the first things which will enter the cvs after .20 so it won't be a long wait for official support

[Gargantua]

Thank you for your very clarifying answer, much appreciated. Will go for 1.6.19 instead then, until the new improvements are implemented.

[Ofloo]

No, eggdrop 1.6.20 doesn't have built-in support for SSL. We will finally add full IPv6 and SSL in the next eggdrop version. The current patch for .19 would need some modifications to apply to .20, but I guess somebody will release an update to it soon.

On the other hand, IPv6 and SSL are the first things which will enter the cvs after .20 so it won't be a long wait for official support

Great I can't wait.

[slennox]

Rootie has updated his patch for 1.6.20:

http://www.egghelp.org/files.htm#patches

I'll boldly advertise it on the home page if no one reports any issues in the next day or two

[Gargantua]

Rootie has updated his patch for 1.6.20:

http://www.egghelp.org/files.htm#patches

I'll boldly advertise it on the home page if no one reports any issues in the next day or two

Indeed a nice piece of information, thanks a lot.

[pseudo]

Rootie has updated his patch for 1.6.20:

http://www.egghelp.org/files.htm#patches

I'll boldly advertise it on the home page if no one reports any issues in the next day or two

I'd recommend the author to run autoconf on this and resubmit it. Otherwise configure warns about --with-ssl. In general, the configure script should be updated by running autoconf on the updated input files, not only by applying an old patch to it.

Code: Select all

configure: WARNING: unrecognized options: --with-ssl

In addition, a user in a help channel reported that the patch didn't recognize OpenSSL. Telling him to run autoconf and then ./configure again fixed the problem.

[slennox]

Rootie sent me this a couple of weeks ago and asked me to post it. Sorry about the delay, have fallen a bit behind with e-mails.

finally that autoconf issue hits me ... I've been aware of that issue
before. The main problem is that autoconf - when applied on a clean eggdrop
source - already creates a completely different configure file (unless I
made something wrong...). Blame the devs for that

What I've done in the previous patches was indeed patching both .ac files
and the main configure file with the patch. My linux system automatically
recognizes OpenSSL - hence I've never tested the new switch "--with-ssl" on
the latest patch. Blame me for that When fixing that issue I've found
another problem that has been present in all previous patches keeping the
code from linking together at the end. I've changed that part in the latest
patch attached to this mail.

He's hoping others will continue to develop the patch to fix any further issues. The updated version of the patch (dated 8 Aug 2010) is here:

http://www.egghelp.org/files/patches/eg ... e.patch.gz

[pseudo]

finally that autoconf issue hits me ... I've been aware of that issue
before. The main problem is that autoconf - when applied on a clean eggdrop
source - already creates a completely different configure file (unless I
made something wrong...). Blame the devs for that

This guy obviously doesn't understand how autotools and autoconf in particular work. Don't blame this on us. The configure script isn't written manually, it's generated by the input templates. You don't edit it, don't patch it, don't do anything to it. You edit it's source files (configure.ac and aclocal.m4 for eggdrop) and run autoconf, which by it's sole purpose will create a completely new file, overwriting any existing old one. Now, since the end user shouldn't have to run or even install autoconf, you distribute the ready file in your patch, in the form of a diff to the old one.

I don't use this patch and actually discourage it's usage, but from a brief look at the new version, the crash-causing problem with the symbols lookup table hasn't been solved. No need to look further.

[fabius]

finally that autoconf issue hits me ... I've been aware of that issue
before. The main problem is that autoconf - when applied on a clean eggdrop
source - already creates a completely different configure file (unless I
made something wrong...). Blame the devs for that

This guy obviously doesn't understand how autotools and autoconf in particular work. Don't blame this on us. The configure script isn't written manually, it's generated by the input templates. You don't edit it, don't patch it, don't do anything to it. You edit it's source files (configure.ac and aclocal.m4 for eggdrop) and run autoconf, which by it's sole purpose will create a completely new file, overwriting any existing old one. Now, since the end user shouldn't have to run or even install autoconf, you distribute the ready file in your patch, in the form of a diff to the old one.

I don't use this patch and actually discourage it's usage, but from a brief look at the new version, the crash-causing problem with the symbols lookup table hasn't been solved. No need to look further.

Completely useless comment! This guy keeps porting the SSL patch to the most recent version of Eggdrop and shares it with the community for free. I've been using his SSL patches for years and I don't have any problems even with the newest version. Feel free to correct his patch if you want. Feel free to make a new SSL patch if you want. But don't blame Rootie for his effort in lifting code. As long as it works on HIS system he's ok with it and he gives his final patch to us so we can use SSL, too.

Just my 2 cents,
fabius

[slennox]

It's a time worn argument whether to release something that's been hacked together in a far from ideal way, or better to have nothing until someone who really knows what they're doing creates something that more people would consider up-to-standard. Users tend to favour the former and developers the latter

Rootie didn't specifically ask to quote this part of his e-mail but I should add it for context:

You can post the text above in the forum to inform users that there's a
"silent update" on the patch that might fix their problems. Actually I
lifted this patch to the newest versions for my private use only. The
patches you get from me work fine on my shell. If it doesn't work on other
systems then everyone can of course change/update the patch. It's open
source

As the patch will become obsolete with the next eggdrop version (in 2 years?
) I've decided not to put any more effort on it. As I said: It works
perfectly fine for me - I can live with it until eggdrop natively supports
SSL.

And here I have to clarify the "in 2 years" remark is not a dig at the pace of eggdrop development, but a little joke in response to a comment I made about the regular timing of his patches. The perils of playing the go-between Rootie can sign up and speak for himself if he has anything to add.

[pseudo]

This guy obviously doesn't understand how autotools and autoconf in particular work. Don't blame this on us. The configure script isn't written manually, it's generated by the input templates. You don't edit it, don't patch it, don't do anything to it. You edit it's source files (configure.ac and aclocal.m4 for eggdrop) and run autoconf, which by it's sole purpose will create a completely new file, overwriting any existing old one. Now, since the end user shouldn't have to run or even install autoconf, you distribute the ready file in your patch, in the form of a diff to the old one.

I don't use this patch and actually discourage it's usage, but from a brief look at the new version, the crash-causing problem with the symbols lookup table hasn't been solved. No need to look further.

Completely useless comment!

O'rly? My comment is useless and your rant isn't? Let's take the facts: the patch for 1.6.20 has serious flaws, which I've pointed out. People do report issues both here in the forum and on IRC support channels. Often these people don't even mention the patch and blame eggdrop itself. So we (the devs) have to check their bugreports only to find that the cause is yet another bug in the ssl patch.
It's naive to believe that since you in particular have no problems, it will work flawlessly for other people. In my 'useless' comment, I pointed out that the most serious bug of the patch stays unresolved. It does crash the bot in many situations. If it worked for you, maybe it would be more useful to stay silent and enjoy using it, wouldn't it? As in difference from you, I've put some effort investigating users' issues with the patch, so I definitely have the grounds to write these comments.

Feel free to correct his patch if you want. Feel free to make a new SSL patch if you want.

Guess what? That's what I've already done. In less than a month 1.8.0 CVS will have complete SSL support. Now I understand the need for a patch before official support is ready, but it's poor quality and has issues. What I'm doing is shedding some light on what are the problems, so why are you even complaining??

But don't blame Rootie for his effort in lifting code.

Sorry, it was not me blaming anybody, it's him blaming us (eggdrop devs) for his own ignorance:

The main problem is that autoconf - when applied on a clean eggdrop source - already creates a completely different configure file (unless I made something wrong...). Blame the devs for that Wink

If anybody is feeling offended by people pointing out his mistakes, it's his own fault.

[pseudo]

It's a time worn argument whether to release something that's been hacked together in a far from ideal way, or better to have nothing until someone who really knows what they're doing creates something that more people would consider up-to-standard. Users tend to favour the former and developers the latter

Rootie didn't specifically ask to quote this part of his e-mail but I should add it for context:

You can post the text above in the forum to inform users that there's a
"silent update" on the patch that might fix their problems. Actually I
lifted this patch to the newest versions for my private use only. The
patches you get from me work fine on my shell. If it doesn't work on other
systems then everyone can of course change/update the patch. It's open
source

As the patch will become obsolete with the next eggdrop version (in 2 years?
) I've decided not to put any more effort on it. As I said: It works
perfectly fine for me - I can live with it until eggdrop natively supports
SSL.

And here I have to clarify the "in 2 years" remark is not a dig at the pace of eggdrop development, but a little joke in response to a comment I made about the regular timing of his patches. The perils of playing the go-between Rootie can sign up and speak for himself if he has anything to add.

I'm not trying to attack Rootie personally, it's good that he shared his work with the others. But if it has serious problems, I have to say it, can't keep silence about this. Would it make sense if I feel offended when people find bugs in eggdrop?

[spithash]

On eggdrop1.6.20... with TCL 8.5 and 8.6

ok so I tried the 1st patch and now I tried this one too:

http://www.egghelp.org/files/patches/eg ... e.patch.gz

SAME error again, the bot connects with SSL and everything, BUT, when you try to log in the party line, you get this wonderful cherry:

Code: Select all

[15:25:34] <nagger> [15:25:34] * Last context: tclhash.c/710 [Tcl proc: *chon:console:chon, param:  $_chonof1 $_chonof2]                                                                      
[15:25:34] <nagger> [15:25:34] * Please REPORT this BUG!                                                                                                                                      
[15:25:34] <nagger> [15:25:34] * Check doc/BUG-REPORT on how to do so.                                                                                                                        
[15:25:34] <nagger> [15:25:34] * Wrote DEBUG
[15:25:34] <nagger> [15:25:34] * SEGMENT VIOLATION -- CRASHING!

I tried in both FreeBSD and OpenBSD.

Epic Fail.

[pseudo]

That's the crash causing bug I was speaking about. This patch shifts down the function lookup table for modules, but doesn't update properly the indices in module.h so sometimes completely wrong functions get called, causing crashes. (These are at the end of the lookup table, including console_chon which is usually called when you log in the partyline.)