On Thu, 2003-05-29 at 15:39, Stas Bekman wrote:
Philippe M. Chiasson wrote:
On Thu, 2003-05-29 at 15:14, Stas Bekman wrote:
I'm not sure why that code was there. So I'm not so confident on removing it. mp1's test suite is too poor for regression testing.
Yeah, you can say that again. I dunno if it would be eventually worth it to port mp1's tests to Apache::Test and try and achieve better covereage. But once again, how long are we planning mp1 will stick around ? Would it be worth it ? (not for 1.28 for sure)
mp1 is here to stay for a while. However if you don't change its guts you are safe from breaking things. Millions of installations are a good test suite ;)
Yeah, but this patch is a good example.
I want to stick with : "If it aint broken, don't fix it"
Only problem, it's broken. But it's hard to have any confidence in patches like this one that touches 'guts' stuff.
I am just worried that these patches will stay out there for quite a long time before they are eventually included.
As I am processing STATUS, I am hoping I'll be able to at least produce a viable patch for most issues (or dismiss them). But on what basis should we decide what does go in a 1.28-tobe and what stays a patch for users to test?
I think previous it worked as follows: patches were applied long before the next release was done, so those who use the cvs version get to test things for a while (1.0 releases aren't very frequent). Based on this, I'd suggest the following approach: if you are confident that a patch won't break things go ahead and commit it into 1.28-tobe. If not, release 1.28 and immediately commit those questionable patches, into 1.29-tobe. You kill 2 birds with this approach:
1) you get the issues resolved without risking breaking things
2) those who need those patches, can always use cvs ;)
This does make a lot more sense, actually. I should have figured that out on my own.
But I then have a question. I was trying to review the current STATUS and at least give each issue a shot. Either submit a patch, get more information, document the problem, etc. And I was trying to get all that done prior to a 1.28 release.
In light of all this, wouldn't it make more sense to ship 1.28 as CVS stands right about now ?
Once it's out there, I could finish my planned review of known issues and get people a change to try it out with CVS.
So I guess I am wondering if there are any issues that would NEED to be resolved before 1.28 could be good enough to be shipped out ?
That sounds right. That's why I've suggested to poll the list if there are any urgent fixes that should go in.
And hey we don't have to wait another year to release 1.29 ;)
p.s. give a try to the latest blead-perl, as i reported earlier modules.t is failing for me.
p.p.s also need to decide whether we bundle Apache::Test with mp1 or not.
__________________________________________________________________ Stas Bekman JAm_pH ------> Just Another mod_perl Hacker http://stason.org/ mod_perl Guide ---> http://perl.apache.org mailto:[EMAIL PROTECTED] http://use.perl.org http://apacheweek.com http://modperlbook.org http://apache.org http://ticketmaster.com
--------------------------------------------------------------------- To unsubscribe, e-mail: [EMAIL PROTECTED] For additional commands, e-mail: [EMAIL PROTECTED]
