[typo] Recent Multi-Blog update Breaks

Piers Cawley pdcawley at bofh.org.uk
Mon Mar 20 07:46:27 EST 2006

Gary Shewan <gpsnospam at gmail.com> writes:
> On 19 Mar 2006, at 18:47, Piers Cawley wrote:
> Thanks for that Piers, it's appreciated because you didn't have to  
> outline all that at all.  Much better attitude than some development  
> communities where any questions are met with "It's open source, we  
> give our time for free, deal with it or disappear"

Well, you know, I was tempted, but I thought of the three virtues of a
programmer (Laziness, Impatience and Hubris). The whole 'deal with it
or disappear' attitude is more along the lines of arrogance than
Hubris. Operating with those virtues tends to mean that stuff goes in
the trunk. If it turns out to be misguided, it gets rolled back. (For
instance, I just rolled back the change that replaced the global
config method with a this_blog method; I had temporarily forgotten
that there are more themes out there than Azure and they all of 'em
use config, not this_blog. Oops.)

> I understand that it needs to be fun, and I've never met a developer  
> that really enjoys a good bug hunt over new functionality ... think  
> I'd have a heart attack if I did ;) 

I dunno, there's a lot to be said for squashing a good bug. When a bug
comes up in the trac that I can fix, I set to it with a will. Problems
arise when I'm either not convinced that something's a bug or I don't
have the knowledge or software to test it (the IE 6 issues, for
instance), or when the defect is obvious, but the design insight
needed to sort it out just isn't forthcoming. Then I go and work on
something that I know how to do.

> There is a growing interest in Typo, but until there is a 4.0
> release I never really recommend it to anyone unless they feel
> confident about it. 

The "Typo installation is as friendly as a cornered rat" issue again
eh? From some of the things he's been saying about this on IRC, I'm
pretty sure you'll like Scott's work on this.

> Everybody I know who isn't too technically minded - and has tried
> Typo - has pulled their hair out in frustration.  I've been bitten
> hard a couple of times with migrations, but that isn't a problem for
> me because I know what working on the edge of trunk means.

And it's because we have to deal with the support issues that arise
when these things happen that we've been working on making the last
few migrations as painless as possible. The new BareMigration system
is a massive improvement here and a thousand thanks are owed to
bronson at rinspin for the fabulous patch that introduced BareMigrations
and rejigged all the migrations to use them. I now have a great deal
more confidence that I can make a migration work independently of any
future changes in the behaviour of typo's model objects and that's an
enormous godsend. I'm now pretty confident that you can migrate up to
the bleading edge from any schema version you like, and that's a
massive improvement on the the bad old days of, um... not all that
long ago recently.

> I still have just over a thousand legacy blog comments sitting in an
> sql file that I have to reinclude because they got lost on one
> migration.  But that's what I get if I only check 99% of everything
> after a trunk migration :)

Ow! Ow! Ow! 

I'd suggest copying your production data into a development database
and then, assuming N is the schema version that the comments you have
come from, do:

  $ rake migrate VERSION=N
  $ <appropriate rdbms commandline> < all_those_comments.sql
  $ rake migrate

and test like crazy.

If it works you can take the production version down for a few minutes
and do the same thing with the production database. If it doesn't,
it'd be good to know why no.

> I like the Typo system, I think you fellas are doing great work.
> Just that a few of us went "Woah!" when we thought we saw a large
> bit of new functionality (that not all users would use) get wedged
> in just after talk of an imminent point release.  After all the
> first line in Trac is "Typo is a lean engine..." ;)

I'm not entirely sure how true that still is. Probably more of an
aspiration than a fact on occasions. 

> Keep it fun, but also take some pride in the fact that people are
> using it and as the number of users grow it speaks volumes about
> your work.

Oh yeah; gaining users is good. Admittedly, the whole "More cattle;
more care" thing rears its head, but but the relationship's more along
the lines of O(logN) than O(N), so that's okay. 

Piers Cawley <pdcawley at bofh.org.uk>

More information about the Typo-list mailing list