Closed Bug 47989 Opened 26 years ago Closed 24 years ago

RUN_AUTOCONF_LOCALLY ineffective

Categories

(SeaMonkey :: Build Config, defect, P3)

x86
All
defect

Tracking

(Not tracked)

VERIFIED FIXED
Future

People

(Reporter: mkaply, Assigned: cls)

Details

Attachments

(1 file, 1 obsolete file)

I put this as OS/2, but it is a problem on any platform that uses RUN_AUTOCONF_LOCALLY. RUN_AUTOCONF_LOCALLY is basically broke. It has two main problems. 1. autoconf is not always run locally. All RUN_AUTOCONF_LOCALLY does it create a dependency to build configure from configure.in. The problem is that this dependency doesn't usually kick because a configure is automagically created by a chron exec and is generally later then configure.in. The only time it kicks is between the time a configure.in change is made and the new configure is checked in. 2. Because we are creating are own configure, we can get conflicts in the configure file on the pull. This prevents the second phase from running completely. I believe there are a few ways to fix this. 1. Have RUN_AUTOCONF_LOCALLY delete the configure before and after the pull. 2. Create a new variable that causes configure to be deleted before and after the pull. Obviously these are both inefficient because autoconf is done each time. 3. Create some new dependency on configure.in such that autoconf is run locally whenever configure.in changes, with no dependency on the actual configure file. I have no idea how this would be done. Anyway, I would be more than happy to work on this if we could find the right way to implement it.
Any luck with setting .cvsignore as leaf suggested?
Status: NEW → ASSIGNED
.cvsignore does work. I came up with another idea. What about adding configure to .cvsignore automatically if RUN_AUTOCONF_LOCALLY is set? I still think there should be something that happens to cause configure to be ignored if RUN_AUTOCONF_LOCALLY is set.
I would just like to point out that both the IRIS machines suffered from this problem over the weekend. Well I would love for this bug to just go away, I still think we need an automated process to add configure to .cvsignore if RUN_AUTOCONF_LOCALLY is set. Maybe just a simple grep on configure in .cvsignore and if not, pipe it out. Please consider it. If not, just close this bug.
OK, I know I am beating a dead horse here, but because the two IRIX machines on ports are again sitting with conflicts in configure, I am at least moving this off of the OS/2 books. Please see my previous comment for a suggested way to fix this automatically or just close this puppy.
OS: OS/2 → All
Just for the record, the IRIX boxes in ports are due to conflicts in nsprpub/configure which this option doesn't even touch. Instead of trying to do something fancy with .cvsignore (which I believe will cause trouble over time), what would be the side effect of always running autoconf after a pull?
I don't think it's a big deal that autoconf will get run every time. The problem is that the build dies on the checkout. So configure would have deleted before the pull if RUN_AUTOCONF_LOCALLY is set to avoid the conflict in the first place.
Setting milestones to Future.
Target Milestone: --- → mozilla0.9
Target Milestone: mozilla0.9 → mozilla0.9.1
Target Milestone: mozilla0.9.1 → Future
Attachment #12973 - Attachment is obsolete: true
Comment on attachment 56375 [details] [diff] [review] Remove configures before checkout & run autoconf after checkout r=mkaply looks good. Do we want to document somewhere that if you want to keep this from happening repeatedly that you unset RUN_AUTOCONF_LOCALLY after the initial pull/build?
Attachment #56375 - Flags: review+
I don't think the note is necessary. The option isn't advertised so a developer would have to look at client.mk to discover it in the first place. And you probably want this to be set always so that you get the configure.in changes when they occur. Right now, having it set means that configure is always regenerated and re-run but the re-running of configure shouldn't change anything unless you edited autoconf.mk by hand or something. Patch checked in.
Status: ASSIGNED → RESOLVED
Closed: 24 years ago
Resolution: --- → FIXED
This is working great for us.
Status: RESOLVED → VERIFIED
Product: Browser → Seamonkey
You need to log in before you can comment on or make changes to this bug.

Attachment

General

Created:
Updated:
Size: