[Gambas-user] "New" gambas-basic.org website

Adrien Prokopowicz adrien.prokopowicz at gmail.com
Mon Feb 12 01:47:19 CET 2018


Le 11/02/2018 à 23:25, Christof Thalhofer a écrit :
> Am 10.02.2018 um 18:45 schrieb Adrien Prokopowicz:
>> That's pretty cool! The only missing feature that I'm used to is having
>> some form of horizontal scalability (with some load balancing), but as
>> you said in an earlier message, I don't think we should be too worried
>> about load. ;-)
> 
> If we generate real load, I think we could pay it then. The only thing
> that creates much traffic IMO are the downloads and they are on Gitlab.

Would there be any limit on the bandwidth for these servers ?

I'm not concerned at all about CPU or disk load. If your servers have 
more than 500KiB of RAM, then all of it would be in filesystem cache, so 
it should be loaded and sent over preeeeetty fast. ;-)

>>>>> It is already there :-) ... in the repo :-) ... in my repo :-)
>>>>
>>>> Can you give me a link ? I searched for "Christof Thalhofer" on GitLab
>>>> but with no success …
>>>
>>> No, the repo is on my computer ;-) that was a joke.
>>>
>>
>> Oh, I didn't get that, sarcasm doesn't get transcribed very well on text
>> (which is why /s was invented I guess).
> 
> It was no sarcasm, because it's real. *Git is decentralized*. Gitlab is
> just one of a lot of places, were the repository really is.
> 
> That is good and bad. The goods thing is, that if Gitlab collapses, the
> repo is still there. On many computers. You literally cannot destroy it!
> 
> The bad thing is: You should not change the history of the repo because
> it is not your own any more. Look at that:
> 
> https://git-scm.com/book/en/v2/Git-Tools-Rewriting-History
> 
> And there: "The Nuclear Option: filter-branch"
> 
> Cite:
> 
> "The command is filter-branch, and it can rewrite huge swaths of your
> history, so you probably shouldn’t use it unless your project isn’t yet
> public and other people haven’t based work off the commits you’re about
> to rewrite."
> 
> So: Do only rewrite history that you have not committed. Never rewrite
> history that is public! And do not store passwords or private keys in
> the repo.
> 
> You can fuckup a lot of other's work by changing history that others
> depend on ... !
> 
> This is different from SVN!

So it wasn't really a joke then. ;-)
Of course, rewriting the history is basically breaking the commit chain, 
and these things tend to not go smoothly for other collaborators. :-)

>> That's great !
>> Do you think it is possible to have an automatic deployment system
>> similar to what GitLab does, so the website could be generated and
>> deployed by simply pushing to the master branch ?
> 
> I see no problem here. Anything that works on a homedir of any debian
> system works here also. We could place a Git repo there and do some code
> in a hook to deploy the website.

Cool. :-)

> Are you the one who is responsible for the current website of Gambas?
> Then please send me a private mail with a public ssh key.

I am not. Benoît is pretty much in charge of everything, I am just 
helping out now and then. ;-)

>> The only problem for now is that the HTTPS site doesn't really work,
>> since the inner frame loads gambaswiki.org in plaintext, and browsers
>> (rightfully) block mixed-content requests. Fixing it requires deploying
>> HTTPS on gambaswiki.org, which we should do at some point, but it's not
>> too much of a deal for now.
> 
> This is a very ugly construction and can eventually lead to downranking
> in search engines.

I am pretty sure it already affects ranking, but yeah it is something 
that should get fixed soon, especially considering wiki/bugtracker 
passwords are being sent in plaintext over the wire …

(Firefox is actually shouting at me for this every time I log in the 
bugtracker or wiki).

> At HS we install free https certificates automatically from
> https://letsencrypt.org/.

Let's encrypt is the best. I set it up on pretty much every single 
project of mine (including the Gambas Playground), and it's quite 
awesome! :-)

-- 
Adrien Prokopowicz


More information about the User mailing list