httpd-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Aram Mirzadeh <>
Subject Re: Patches patched... or something
Date Fri, 01 Mar 1996 20:27:10 GMT
> On Fri, 1 Mar 1996, Paul Richards wrote:
> Not quite. In fact, if it was just that, I might have patched it in. The 
> actual case is more complex: A week ago, when I was patching the patches 
> together just to see if they would, I found that the patch in question 
> did not work. There was a missing #endif, but more than that, both Rob 
> Hartill and I said that <paths.h> did not exist on HP-UX, at least.
> So it's not that Aram didn't know that HP didn't have the file, he did 
> (assuming he read the mail). That's my real concern: that a problem was 
> pointed out, but was not addressed.

Actually, no I must have missed that one...  I get 300-400 emails a day, not
sure if that's too much for everyone else or not, but it's just about enough
for me.  So sometimes I just can glace at an email before discarding it, 
specially if it's a mailling list.. but if you send it to me specificly
it always get's read.  I'm not sure what the solution here is, but I'm sure
what we're doing isn't the right one.  I don't think a patch should be 
vetoed, or not used because it doesn't work on a first glace.  I thought the
whole idea behind going to CVS would be that the work around of the same
patch would go faster, than I sumbit a patch, it fails on 2 platforms, I
resumbit it, it stops working on another platform, and then 3rd sumbittion 
of the same patch works for all.... this would have taken 2 weeks on the 
previous voting system.  With cvs it should be I sumbit a patch a coupel of
days go by, we stop for a couple of days for everyone to checkout the latest
test, and apply any patches they need for their system.  Then vote on the
entire disterbution, and see if anything needs to be taken out.  

If I'm mistaken at how this is supposed to work, would someone please let 
me know in a step by step how its supposed to work?

> Oh, another CVS question: what are we doing about CHANGES? Some of the
> people committing stuff today have updated CHANGES, some haven't. When I
> put in the patches from for_Apache_1.1b0, I didn't, on the assumption that
> the release builder (whenever a release gets made) would look at the RCS
> logs and generate it from that. In fact, it could probably be generated
> very simply. In fact, just taking the apache-cvs mail log and taking out
> the SMTP headers would very nearly give you a complete changelog. 

Don't you think we should only incorporate the major changes in to the 
/CHANGES file?  Each small patch should only update the headers of the 
file they're updating.. no?

> So the question is: what is procedure here? I vote for some sort of 
> automated, rlog-based approach. It just makes life easier, yes?

Good question.
#define BITCOUNT(x)     (((BX_(x)+(BX_(x)>>4)) & 0x0F0F0F0F) % 255)
#define  BX_(x)         ((x) - (((x)>>1)&0x77777777)                    \
                             - (((x)>>2)&0x33333333)                    \
                             - (((x)>>3)&0x11111111))

                -- really weird C code to count the number of bits in a word
Aram Mirzadeh

View raw message