On Aug 1, 2018, at 3:43 PM, John P. Rouillard <rouilj+fos...@cs.umb.edu> wrote:
> In message <23ebb036-9e71-4e02-b496-1361e3852...@etr-usa.com>,
> Warren Young writes:
>> On Jul 31, 2018, at 7:47 AM, Richard Hipp <d...@sqlite.org> wrote:
>>> The intent is to replace this mailing list
>> Are you also going to import all the old content, so that people can
>> clone the forum repo — assuming you still plan to keep the forum and
>> code as separate repos — and get a locally-searchable archive?
> According to the comment above

Not according to me, but to drh:


> I must have missed that if it was stated.

The current forumtest1 contents are currently small enough to read 
front-to-back in 15 minutes or so. ;)

> Since I clone the fossil repo to my RasPI I am really not interested
> in cloning the forum/mailing list along with the source code.

If you won’t use the mailing list archive, fine, I have no particular wish to 
force you to clone it.  

However, when weighing your wishes, we must keep the costs in perspective: 
According to that same post, we’re only talking about 35 MB or so of data.

The Raspbian Desktop SD card image is currently about 4.8 GB unpacked, so with 
the smallest-possible 8 GB card, a SQLite mailing list archive clone is small 
enough to rattle around in the free space like a BB in a milk jug.

If you’re using Raspbian Lite, you can install on a card as small as 2 GB, and 
you’ll have enough space left over for some development tools, Fossil, and 
several repository clones, including those of the Fossil and SQLite mailing 

What I’m saying here is, give me a use case where 35 MB actually matters here 
in 2018.

I can give one: those behind SQLite and Fossil might not want to pay the higher 
bandwidth bills resulting from a repository that’s 35+ MB larger.  At their 
scale, it might actually matter, and in that case, they should calve it off as 
a separate repository so that only those that actually want to make use of it 
will bother to clone it.  (I do, and I will, if allowed.)

Here’s a thought for you: how many old forums (generic definition) have you 
been part of in the past that have since disappeared, taking their archives 
with them into oblivion?  Wouldn’t you like to have a copy of some of that?  
For me, the answers are “dozens,” and “yes”.  The Fossil Forums Feature gives 
us that, finally.

> I throw away 99+% of the traffic on the mailing list after reading
> it.

You also probably ignore 99+% of past software versions, 99+% of past checkin 
comments, 99+% of old wiki content…  The thing a VCS should provide is access 
to the odd 0.01% of past artifacts that you can’t identify in advance of need.

If I’m writing a new ticket for a bug that was discussed on the mailing list 
months ago, I’d sure rather refer to it using a link to a forum post in the 
same repository than to some third-party mail archive service that has a fair 
chance of changing their URL scheme or even going out of business before the 
ticket is closed.

> A method that allows syncing without forum artifacts or a command to
> purge forum artifacts would be helpful.

A method to get some kind of thinned-out clone has been a longstanding wish.

I think the main problem is that it’s a lot of difficulty for a fairly uncommon 
use case.  Most of the time for most projects — including, critically, SQLite 
and Fossil-SCM.org — a full clone is cheap enough that it just isn’t worth 
doing the work.
fossil-users mailing list

Reply via email to