[squeak-dev] Refactoring browser loading

tim Rowledge tim at rowledge.org
Thu Jan 17 05:23:09 UTC 2019



> On 2019-01-16, at 6:45 PM, Chris Muller <asqueaker at gmail.com> wrote:
> 
> Hi Tim,
> 
>> Trying the SqueakMap approach hasn't been much more successful;
> 
> RefactoringTools on SqueakMap works.

As I said, indeed it does, once the loading has been handled. I've been looking at the code in SMSqueakMp>>loadFullFrom: with a view to fixing the lack of error handling; and as is so often the case it leads to many other places where there is no checking to see if the response to an 'http get' has problems, so it will take a little more thought.

> 
>> after allowing for not-yet-updated packages,
> 
> The default filter shipped in a 5.2 image allows you to see what is
> updated and working at a 5.2 level.

But not in a current development image, where I have to clear the filter. Which is obvious once I've remembered it, after not having had to think about it in some time.

> 
>> I see several 'RefactoringBrowser xxx' entries, all very old. A 'Refactoring Tools' entry looks more interesting (though confusing name differences can make using searches such fun, always worth trying to remember)
> 
> Right, and also worth remembering that this is simply the result of it
> being a model of Squeak's history.  Turn back on the filter and all
> that "old" noise will go away.
> 
>> and starts to load.
>> 
>> But then it fails because SMSqueakMp>>loadFullFrom: tries to fetch some zipped data and didn't check the returned info - which in this case is a 504 error notification. Oops...
>> Restarting the method seems to make it happier...

As above, the problem is that there is no checking for even the possibility of the server not returning the desired content. So we don't get to (maybe) compensate for the elderly nature of the server or any other problem. We can do better with some thought.


>> twiddle, twiddle, ooh, could not access 'http://www.squeaksource.com/rb' in MCHttpRepository>>#httpGet:arguments: Restart *that* and continue...

I'm not sure what the actual error was there, other than a general 'could not access'. Maybe something in the socket code could work out a more helpful error so that, again, we could try to do something helpful. Net services go missing, or connections fail, or dns wotsits get mangled and so forth. I bet we can do something helpful in most cases.

I'm pleased that there is a version that loads and seems to do good things. The refactoring stuff in the VW8.3 image is about the only 'new' thing I have been pleased to find whilst using that system; amazingly it's well over 20 years since I last used VW. 

tim
--
tim Rowledge; tim at rowledge.org; http://www.rowledge.org/tim
Useful random insult:- He's not a complete idiot -- some parts are missing.




More information about the Squeak-dev mailing list