Closed Bug 227078 Opened 21 years ago Closed 21 years ago

build should not require File::Spec version 0.8

Categories

(SeaMonkey :: Build Config, defect)

defect
Not set
normal

Tracking

(Not tracked)

RESOLVED FIXED

People

(Reporter: dbaron, Assigned: ian)

References

Details

(Keywords: fixed1.6)

Attachments

(1 file, 1 obsolete file)

The build should not require File::Spec version 0.8. It has proven to be a pain and there's no good reason why we need it. See bug 184182 and bug 224557. I think we should remove this requirement for 1.6b, either by fixing the XUL preprocessor (as in the patch in bug 184182) or by not using it in Seamonkey until the XUL preprocessor is fixed.
Flags: blocking1.6b?
I have to admit being confused as to why you wouldn't have File::Spec 0.8. All the machines I use have File::Spec versions greater than 0.8. Heck even DreamHost has File:Spec 0.83, and DreamHost are notorious for being slow to update, and they use Debian, who are themselves notorious for being years behind the rest of the world. And they've had it for a good year. Machines without File::Spec 0.8 or higher really ought to join us in the 21st century. I personally can't do much about this since I can't actually find a machine that _doesn't_ have File::Spec 0.8 or above, so I couldn't test a patch that made us not dependent on 0.8. Someone wrote a patch for this in one of the bugs, but they didn't respond to my review question.
So take that patch, modify it so you're happy with it, and assume it still doesn't require 0.8. It's not hard. It should take you about an hour to do it and get it reviewed, super-reviewed, and checked in, which is less time than I spent upgrading the tinderboxes that needed the upgrade. Never mind all the time this will force people to spend who want to build the 1.6 release. There are good reasons to upgrade the build requirements, but "I'm too lazy to spend a few minutes fixing up a patch and getting it checked in" isn't one of them.
(Note that Hixie wasn't the one responsible for foisting this build requirement change on Seamonkey, and the people I'm really annoyed at are those who did.)
I don't understand the change that I asked about in the patch, and I have a feeling that it is required to not depend on 0.8, although I have no way of verifying this. Hence my reluctance to accept it or check it or a variant in myself. Trying to work out what changes would be needed without any more than the module's changelog would take time, and at the moment I do not have cycles to spend on it. Sorry. (I also find it quite amusing that a tool I was writing for fun somehow ended up requiring review, super-review, and my staying up for an hour checking tinderboxes every time I change it.)
Axel, can you do what's necessary to get this in?
My phd defense is on next monday, and I don't have perl versions ancient enough at my disposal, sorry. Well, I have a perl5.00503, but that doesn't seem to have File:Spec:Unix, though it has a (empty) dir File/Spec. No idea why that would be, it could be the admins of that machine. Doesn't seem like I could review http://bugzilla.mozilla.org/attachment.cgi?id=129428&action=edit, either. Sebastion (or Alan?) should prolly answer Hixie's question.
moving nomination to 1.6 final. Not going to hold beta for this.
Flags: blocking1.6b?
Flags: blocking1.6b-
Flags: blocking1.6?
Attached patch patch to remove File::Spec 0.8 (obsolete) — Splinter Review
This patch was created/tested on a FreeBSD 4.9 system with perl 5.005_03 and File::Spec 0.6. Each of the FreeBSD ports patches for Mozilla and Firebird and <http://bugzilla.mozilla.org/attachment.cgi?id=129428&action=view> is in itself inadequate. This one's a bit more thorough, and some attention was given to making the paste job look not so awful:)
Looks good. I'll look at this in more detail ASAP.
need to make a call in this 1.6 very soon...
Attachment #138659 - Flags: superreview?(bryner)
Attachment #138659 - Flags: review?(bryner)
(r/sr from dbaron would be fine too)
Attachment #138659 - Flags: superreview?(bryner)
Attachment #138659 - Flags: superreview+
Attachment #138659 - Flags: review?(bryner)
Attachment #138659 - Flags: review+
Comment on attachment 138659 [details] [diff] [review] minor changes from attachment 138256 [details] [diff] [review] to fit the local style more a=asa (on behalf of drivers) for checkin to 1.6
Attachment #138659 - Flags: approval1.6+
ok, lets get this in quickly if we are going to do it for 1.6... who can do the deed?
Flags: blocking1.6? → blocking1.6+
This is now in on the trunk. Someone with a 1.6 branch should do the branch checkin, I don't have a tree. (I have a very localised checkout of just the preprocessor.pl file.) Marking FIXED for the trunk.
Status: NEW → RESOLVED
Closed: 21 years ago
Resolution: --- → FIXED
Checked in to 1.6 branch.
Keywords: fixed1.6
Blocks: 184182
This broke ActiveState builds (see bug 232002). I have no idea why. Unless we can work out why, I intend to back this change out.
Product: Browser → Seamonkey
You need to log in before you can comment on or make changes to this bug.

Attachment

General

Creator:
Created:
Updated:
Size: