Sign in Join

Live chat for Free Pascal, Lazarus, Delphi and every Pascal programmer.

Mailing lists / lazarus / 2011-12

Cannot build current trunk - codetools

13 messages · 7 people · –

Reply by email Subscribe to this list Your subscription

Read-only archive. Use j and k to move between messages.

zeljko
to zeljko
zeljko wrote 8 lines
Compiling codetoolsstructs.pas codetoolsstructs.pas(101,5) Error: VAR and TYPE are allowed only in generics codetoolsstructs.pas(107,5) Fatal: Syntax error, "=" expected but "identifier FCOMPAREKEYITEMFUNC" found Fatal: Compilation aborted svn r. 33928 fpc-2.4.5

Forgot to mention, 33925 builds fine, so Mattias is suspected ;)

zeljko
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.lazarus.freepascal.org/pipermail/lazarus/attachments/20111203/7a81c540/attachment.html>

Reply by email

Frank Church
to Mattias Gaertner
Mattias Gaertner wrote 11 lines
On Sat, 3 Dec 2011 16:15:24 +0100 Bart wrote: > Reported by lainz in the bugtracker: > http://bugs.freepascal.org/view.php?id=20799 Please test. Mattias

I don't know how welcome or apt this comment may be as it appears to a contentious issue, but it has been on my mind for some time.

Every now and then I see a message about trunk not compiling or some the feature being broken.

I thought one of the reasons for proposing a change to Git or some other distributed VCS is that you do all your work in your local branch, merge with HEAD frequently and you only push your changes to the main repository if everything works properly.
It is understood that features may get broken as it they are hard to test, but shouldn't compilation succeed? This suggests work is being 'saved' to the repository which shouldn't be the case.

I am not proposing a change to git or other if core developers are content with Subversion, I'd just want to know if it is a reflection of working practices which may or may not be influenced by the choice of VCS and methodologies surrounding its usage.

Show quoted text 6 lines
-- _______________________________________________ Lazarus mailing list http://lists.lazarus.freepascal.org/mailman/listinfo/lazarus
Frank Church ======================= http://devblog.brahmancreations.com -------------- next part -------------- An HTML attachment was scrubbed... URL: <http://lists.lazarus.freepascal.org/pipermail/lazarus/attachments/20111203/01525b8b/attachment.html>

Reply by email

Martin
to Frank Church
Frank Church wrote 11 lines
I don't know how welcome or apt this comment may be as it appears to a contentious issue, but it has been on my mind for some time. Every now and then I see a message about trunk not compiling or some the feature being broken. I thought one of the reasons for proposing a change to Git or some other distributed VCS is that you do all your work in your local branch, merge with HEAD frequently and you only push your changes to the main repository if everything works properly.

How would that help?
I guess Mattias was using a newer FPC (2.5 or 2.7), so compilation worked for him, and he would have pushed.

I had that issue myself a few times (either FPC version or some local $DEFINE / bgt it may also be caused by platform differences or a installed package).

To solve this each devel would have to compile (and test) the entire IDE with a variety of settings/FPC ...

Show quoted text 9 lines
It is understood that features may get broken as it they are hard to test, but shouldn't compilation succeed? This suggests work is being 'saved' to the repository which shouldn't be the case. I am not proposing a change to git or other if core developers are content with Subversion, I'd just want to know if it is a reflection of working practices which may or may not be influenced by the choice of VCS and methodologies surrounding its usage.

I don't think it is a big issue, thinks get usually fixed very quickly.

Reply by email

Mattias Gaertner
to Frank Church: I don't know how welcome or apt this comment may be as it appears to a contentious issue,…
Show quoted text 25 lines
On 3 December 2011 15:48, Mattias Gaertner wrote: > On Sat, 3 Dec 2011 16:15:24 +0100 > Bart wrote: > > > Reported by lainz in the bugtracker: > > http://bugs.freepascal.org/view.php?id=20799 > > Please test. > > Mattias > > I don't know how welcome or apt this comment may be as it appears to a contentious issue, but it has been on my mind for some time. Every now and then I see a message about trunk not compiling or some the feature being broken. I thought one of the reasons for proposing a change to Git or some other distributed VCS is that you do all your work in your local branch, merge with HEAD frequently and you only push your changes to the main repository if everything works properly. It is understood that features may get broken as it they are hard to test, but shouldn't compilation succeed?

It compiled with 2.5.1 and 2.7.1.
I copied the code from a place I thought I had tested with 2.4.5. Apparently I have buggy memory. Sorry for that.

This suggests work is being 'saved' to the repository which shouldn't be the case.

Or it suggests that you need some more imagination. FPC is a very fast compiler, so I don't believe any one is committing without compiling first.

I am not proposing a change to git or other if core developers are content with Subversion, I'd just want to know if it is a reflection of working practices which may or may not be influenced by the choice of VCS and methodologies surrounding its usage.

?

Mattias

Reply by email

Mattias Gaertner
to Martin: How would that help? I guess Mattias was using a newer FPC (2.5 or 2.7), so compilation…
[...] I had that issue myself a few times (either FPC version or some local $DEFINE / bgt it may also be caused by platform differences or a installed package).

Yes, there were some this year.
The last one two days ago, when it compiled everywhere but Linux 64bit.

To solve this each devel would have to compile (and test) the entire IDE with a variety of settings/FPC ...

... which would be diametrically to the policy of committing in small changes.

Mattias

Reply by email

Sven Barth
to Frank Church: I don't know how welcome or apt this comment may be as it appears to a contentious issue,…
Frank Church wrote
Every now and then I see a message about trunk not compiling or some the feature being broken.

Trunk is supposed to be "use at your own risk". There are phases were it's rather save to use trunk, but there are also those where it needs to be considered "experimental". An example for this was FPC's trunk the last weeks since the merge of the cpstrnew branch.

Regards,
Sven

Reply by email

Unknown author
to Sven Barth

Quoting Sven Barth :

Show quoted text 8 lines
On 03.12.2011 18:46, Frank Church wrote: > Every now and then I see a message about trunk not compiling or some the > feature being broken. Trunk is supposed to be "use at your own risk". There are phases were it's rather save to use trunk, but there are also those where it needs to be considered "experimental". An example for this was FPC's trunk the last weeks since the merge of the cpstrnew branch.

+1 with one difference. When we do experimental things in Lazarus, then it's usually ifdeffed , so ordinary users won't be disturbed by new experimental stuff.Lazarus trunk is usually broken (few times in last few weeks) because devels forget
to add some unit or don't build lazarus (and test their changes) before commiting.
Anyway it's fixed in few minutes ... max 1-2 hours, so @Frank: if you run into broken trunk, then go to lazarus viewvc and see where it's broken (or write your last good trunk version - that's what I'm doing), and svn update -r LASTGOODKNOWNTOWORK.

zeljko

Reply by email

Frank Church
to Mattias Gaertner: On Sat, 3 Dec 2011 17:46:49 +0000 It compiled with 2.5.1 and 2.7.1. I copied the code…
Mattias Gaertner wrote 46 lines
On Sat, 3 Dec 2011 17:46:49 +0000 Frank Church wrote: > On 3 December 2011 15:48, Mattias Gaertner < >wrote: > > > On Sat, 3 Dec 2011 16:15:24 +0100 > > Bart wrote: > > > > > Reported by lainz in the bugtracker: > > > http://bugs.freepascal.org/view.php?id=20799 > > > > Please test. > > > > Mattias > > > > > I don't know how welcome or apt this comment may be as it appears to a > contentious issue, but it has been on my mind for some time. > > Every now and then I see a message about trunk not compiling or some the > feature being broken. > > I thought one of the reasons for proposing a change to Git or some other > distributed VCS is that you do all your work in your local branch, merge > with HEAD frequently and you only push your changes to the main repository > if everything works properly. > It is understood that features may get broken as it they are hard to test, > but shouldn't compilation succeed? It compiled with 2.5.1 and 2.7.1. I copied the code from a place I thought I had tested with 2.4.5. Apparently I have buggy memory. Sorry for that. > This suggests work is being 'saved' to > the repository which shouldn't be the case. Or it suggests that you need some more imagination. FPC is a very fast compiler, so I don't believe any one is committing without compiling first.

I am not aiming to be critical as such. For a small of group of core updaters who are up to date on each others work it won't be problem, but occasional users do get bitten every now and then, especially if they need some new features in the LCL and other libraries.

I even thought of adding a note to Graeme not to jump in[?]

Show quoted text 16 lines
> I am not proposing a change to git or other if core developers are content > with Subversion, I'd just want to know if it is a reflection of working > practices which may or may not be influenced by the choice of VCS and > methodologies surrounding its usage. ? Mattias -- _______________________________________________ Lazarus mailing list http://lists.lazarus.freepascal.org/mailman/listinfo/lazarus
Frank Church ======================= http://devblog.brahmancreations.com -------------- next part -------------- An HTML attachment was scrubbed... URL: <http://lists.lazarus.freepascal.org/pipermail/lazarus/attachments/20111204/4ac3b8df/attachment.html> -------------- next part -------------- A non-text attachment was scrubbed... Name: not available Type: image/gif Size: 96 bytes Desc: not available URL: <http://lists.lazarus.freepascal.org/pipermail/lazarus/attachments/20111204/4ac3b8df/attachment.gif>

Reply by email

Sven Barth
to Unknown author: Quoting Sven Barth : +1 with one difference. When we do experimental things in Lazarus,…
Show quoted text 15 lines
Quoting Sven Barth : > On 03.12.2011 18:46, Frank Church wrote: >> Every now and then I see a message about trunk not compiling or some the >> feature being broken. > > Trunk is supposed to be "use at your own risk". There are phases were > it's rather save to use trunk, but there are also those where it needs > to be considered "experimental". An example for this was FPC's trunk > the last weeks since the merge of the cpstrnew branch. +1 with one difference. When we do experimental things in Lazarus, then it's usually ifdeffed , so ordinary users won't be disturbed by new experimental stuff.

I don't not mean this in a way that you include experimental features that are not yet fully implemented, but more in the case of adding a feature which - by accident or whatever - works differently or not at all on a different platform (or doesn't even compile there). That's why I also mentioned FPC's trunk, because after the merge of cpstrnew the Unix platforms where initially more broken than the Windows ones.

Regards,
Sven

Reply by email

Martin
to Frank Church: I am not aiming to be critical as such. For a small of group of core updaters who are up…
Frank Church wrote 10 lines
> I don't know how welcome or apt this comment may be as it appears to a > contentious issue, but it has been on my mind for some time. > > Every now and then I see a message about trunk not compiling or some the > feature being broken. >

....

Show quoted text 7 lines
I am not aiming to be critical as such. For a small of group of core updaters who are up to date on each others work it won't be problem, but occasional users do get bitten every now and then, especially if they need some new features in the LCL and other libraries.

If you are using snapshots or SVN then there are risks. And there is a warning about that.

I know a lot of people use it anyway, and think of it as rather save (and that provides for a lot of well needed extra testing).

But let me remind you: Just a few weeks back several people lost their entire project folder, because of a bug in SVN and trunk snapshots.

Anyone taking that risk, should think of being unable to compile the IDE as a real minor issue.
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.lazarus.freepascal.org/pipermail/lazarus/attachments/20111204/d827fbfa/attachment.html>

Reply by email