RE: [PHP-DEV] RfC: version names
At 17:11 31-1-2003, Sascha Schumann wrote: What is the checkout date here? Let's focus on the real issue. The issue is not which date - it is a means to asses, which source changes broke/fixed anything. Something along this line, is looking at the issue, not the indicator: find . -type f -name *.[chy] -print | xargs grep '/\*[[:space:]]*\$Id:' |sed -e s|^[^:]*:|| CVSIDS This should be in run-tests.php, or a seperate script even, to allow you to track down build breakers. It would be required that all source files, have a conforming header so this has an extra bonus. With kind regards, Melvyn Sopacua ?php include(not_reflecting_employers_views.txt); ? -- PHP Development Mailing List http://www.php.net/ To unsubscribe, visit: http://www.php.net/unsub.php
Re: [PHP-DEV] RfC: version names
On Fri, 31 Jan 2003, [iso-8859-1] Kai Schröder wrote: Hi all, I'm working on a new framework for analyzing results of make test for QA Team (see php-qa list). Talking with Sebastian Nohn, I realized, that we can't differ the versions in detail. That means, that we can't say which source packet was used to build php. Is there any way to name the versions 4.3.0 (release), 4.3.1-dev (taken from cvs, additional timestamp would be nice), 4.3.1-snapMMDDHHii (taken from snaps.php.net) and 4.3.1-win32-snapMMDDHHii (Win32 Snapshot)? I suppose we can have the snapshot generator do that, but it would still hard to differentiate between CVS versions then. Derick -- - Derick Rethans http://derickrethans.nl/ PHP Magazine - PHP Magazine for Professionals http://php-mag.net/ - -- PHP Development Mailing List http://www.php.net/ To unsubscribe, visit: http://www.php.net/unsub.php
RE: [PHP-DEV] RfC: version names
I suppose we can have the snapshot generator do that, but it would still hard to differentiate between CVS versions then. Sounds a little bit stupid, but why the snapshot generator can't simply commit a changed /main/php_version.h every 2 hours? The php-cvs mailer should ignore the commits (ignore all from user vercommt for instance) and there must be a secure way to make commits for this special file without having full dev karma. Regards, Kai -- PHP Development Mailing List http://www.php.net/ To unsubscribe, visit: http://www.php.net/unsub.php
RE: [PHP-DEV] RfC: version names
On Fri, 31 Jan 2003, Kai Schröder wrote: I suppose we can have the snapshot generator do that, but it would still hard to differentiate between CVS versions then. Sounds a little bit stupid, but why the snapshot generator can't simply commit a changed /main/php_version.h every 2 hours? The php-cvs mailer should ignore the commits (ignore all from user vercommt for instance) and there must be a secure way to make commits for this special file without having full dev karma. That can be done, but that means 12 commits a day for a single file. I dont think that's a good idea. Derick -- - Derick Rethans http://derickrethans.nl/ PHP Magazine - PHP Magazine for Professionals http://php-mag.net/ - -- PHP Development Mailing List http://www.php.net/ To unsubscribe, visit: http://www.php.net/unsub.php
RE: [PHP-DEV] RfC: version names
That can be done, but that means 12 commits a day for a single file. I dont think that's a good idea. That's true. Now we have round about 30 commits each day. 12 more will make less than 5.000 blocks (5MB for 1k-blocks) more disk usage per year. Where do you see the real problem if the commits are not mailed to php-cvs list? Regards, Kai -- PHP Development Mailing List http://www.php.net/ To unsubscribe, visit: http://www.php.net/unsub.php
RE: [PHP-DEV] RfC: version names
I suppose we can have the snapshot generator do that, but it would still hard to differentiate between CVS versions then. Sounds a little bit stupid, but why the snapshot generator can't simply commit a changed /main/php_version.h every 2 hours? That can be done, but that means 12 commits a day for a single file. I dont think that's a good idea. Is the ChangeLog updated religiously? If so, CVS builds could include a datestamp acquired from that. James --- Outgoing mail is certified Virus Free. Checked by AVG anti-virus system (http://www.grisoft.com). Version: 6.0.449 / Virus Database: 251 - Release Date: 27/01/2003 This e-mail has been scanned for all viruses by Star Internet. The service is powered by MessageLabs. For more information on a proactive anti-virus service working around the clock, around the globe, visit: http://www.star.net.uk -- PHP Development Mailing List http://www.php.net/ To unsubscribe, visit: http://www.php.net/unsub.php
Re: [PHP-DEV] RfC: version names
Kai Schröder wrote: That's true. Now we have round about 30 commits each day. 12 more will make less than 5.000 blocks (5MB for 1k-blocks) more disk usage per year. Where do you see the real problem if the commits are not mailed to php-cvs list? that is 12 commits per day to a single file after a year this file would have a cvs release id of something like 1.4500 take a look at http://cvs.php.net/cvs.php/php4/NEWS to see why this might be a not-so-good idea ... -- Six Offene Systeme GmbH http://www.six.de/ i.A. Hartmut Holzgraefe Email: [EMAIL PROTECTED] Tel.: +49-711-99091-77 Sie finden uns auf der CeBIT in Halle 6/H44 http://www.six.de/cebit2003/ -- PHP Development Mailing List http://www.php.net/ To unsubscribe, visit: http://www.php.net/unsub.php
RE: [PHP-DEV] RfC: version names
On Fri, 31 Jan 2003, [ISO-8859-15] Kai Schröder wrote: That can be done, but that means 12 commits a day for a single file. I dont think that's a good idea. That's true. Now we have round about 30 commits each day. 12 more will make less than 5.000 blocks (5MB for 1k-blocks) more disk usage per year. Where do you see the real problem if the commits are not mailed to php-cvs list? Cluttering the history of an important file is an extremely bad idea. - Sascha -- PHP Development Mailing List http://www.php.net/ To unsubscribe, visit: http://www.php.net/unsub.php
RE: [PHP-DEV] RfC: version names
On Fri, 31 Jan 2003, James Aylett wrote: I suppose we can have the snapshot generator do that, but it would still hard to differentiate between CVS versions then. Sounds a little bit stupid, but why the snapshot generator can't simply commit a changed /main/php_version.h every 2 hours? That can be done, but that means 12 commits a day for a single file. I dont think that's a good idea. Is the ChangeLog updated religiously? If so, CVS builds could include a datestamp acquired from that. Only daily, but it seems like the correct way to do it to me. Derick -- - Derick Rethans http://derickrethans.nl/ PHP Magazine - PHP Magazine for Professionals http://php-mag.net/ - -- PHP Development Mailing List http://www.php.net/ To unsubscribe, visit: http://www.php.net/unsub.php
RE: [PHP-DEV] RfC: version names
On Fri, 31 Jan 2003, Kai Schroder wrote: Sascha found a simple but good solution: date /CURRENT_TIME This can be done ... for manual checkouts without changing source files in cvs. How is that possible? Derick -- - Derick Rethans http://derickrethans.nl/ PHP Magazine - PHP Magazine for Professionals http://php-mag.net/ - -- PHP Development Mailing List http://www.php.net/ To unsubscribe, visit: http://www.php.net/unsub.php
RE: [PHP-DEV] RfC: version names
Derrick wrote: date /CURRENT_TIME This can be done ... for manual checkouts without changing source files in cvs. How is that possible? I don't think it is, because it needs to be done at checkout time, not at build time. I previously suggested (with Derrick's reply): Is the ChangeLog updated religiously? If so, CVS builds could include a datestamp acquired from that. Only daily, but it seems like the correct way to do it to me. That should be sufficiently fine-grained, at least for the moment. It certainly tightens the range of a codebase in a bug report from weeks or months down to a single day or so. Cheers, James --- Outgoing mail is certified Virus Free. Checked by AVG anti-virus system (http://www.grisoft.com). Version: 6.0.449 / Virus Database: 251 - Release Date: 27/01/2003 This e-mail has been scanned for all viruses by Star Internet. The service is powered by MessageLabs. For more information on a proactive anti-virus service working around the clock, around the globe, visit: http://www.star.net.uk -- PHP Development Mailing List http://www.php.net/ To unsubscribe, visit: http://www.php.net/unsub.php
RE: [PHP-DEV] RfC: version names
Sascha found a simple but good solution: date /CURRENT_TIME This can be done by snapshot generator and for manual checkouts without changing source files in cvs. Than /run-tests.php can send the content of this file with the test results. Regards, Kai -- PHP Development Mailing List http://www.php.net/ To unsubscribe, visit: http://www.php.net/unsub.php
RE: [PHP-DEV] RfC: version names
How is that possible? I don't think it is, because it needs to be done at checkout time, not at build time. What are you smoking? That's a one line addition to the snapshot script. - Sascha, creator, snaps.php.net -- PHP Development Mailing List http://www.php.net/ To unsubscribe, visit: http://www.php.net/unsub.php
RE: [PHP-DEV] RfC: version names
On Fri, 31 Jan 2003, Sascha Schumann wrote: How is that possible? I don't think it is, because it needs to be done at checkout time, not at build time. What are you smoking? That's a one line addition to the snapshot script. For manual checkouts that has nothing to do with snapshots, so he isn't smoking anything. Derick -- - Derick Rethans http://derickrethans.nl/ PHP Magazine - PHP Magazine for Professionals http://php-mag.net/ - -- PHP Development Mailing List http://www.php.net/ To unsubscribe, visit: http://www.php.net/unsub.php
RE: [PHP-DEV] RfC: version names
For manual checkouts that has nothing to do with snapshots, so he isn't smoking anything. echo test -r CHECKOUT_TIME || date CHECKOUT_TIME buildconf would be sufficient for that case. It's not perfect, because it is conceivable that someone does not run buildconf immediately. But that is unlikely from my POV. - Sascha -- PHP Development Mailing List http://www.php.net/ To unsubscribe, visit: http://www.php.net/unsub.php
RE: [PHP-DEV] RfC: version names
On Fri, 31 Jan 2003, Sascha Schumann wrote: For manual checkouts that has nothing to do with snapshots, so he isn't smoking anything. echo test -r CHECKOUT_TIME || date CHECKOUT_TIME buildconf would be sufficient for that case. It's not perfect, because it is conceivable that someone does not run buildconf immediately. But that is unlikely from my POV. It's not actually, as we were just trying to *solve* this problem and prevent getting test results with the wrong date/time. Derick, creator -- - Derick Rethans http://derickrethans.nl/ PHP Magazine - PHP Magazine for Professionals http://php-mag.net/ - -- PHP Development Mailing List http://www.php.net/ To unsubscribe, visit: http://www.php.net/unsub.php
RE: [PHP-DEV] RfC: version names
It's not actually, as we were just trying to *solve* this problem and prevent getting test results with the wrong date/time. So what happens if I do this? cd php4 cvs upd and a day later do this? cd php4/ext/standard cvs upd What is the checkout date here? - Sascha -- PHP Development Mailing List http://www.php.net/ To unsubscribe, visit: http://www.php.net/unsub.php
RE: [PHP-DEV] RfC: version names
On Fri, 31 Jan 2003, Sascha Schumann wrote: It's not actually, as we were just trying to *solve* this problem and prevent getting test results with the wrong date/time. So what happens if I do this? cd php4 cvs upd and a day later do this? cd php4/ext/standard cvs upd What is the checkout date here? I've no idea, do you? I think we just should not allow submissions with test results if they're not made by a snapshot or our phport.sh thingy for automatic testing. Derick -- - Derick Rethans http://derickrethans.nl/ PHP Magazine - PHP Magazine for Professionals http://php-mag.net/ - -- PHP Development Mailing List http://www.php.net/ To unsubscribe, visit: http://www.php.net/unsub.php
Re: [PHP-DEV] RfC: version names
Hi, So what happens if I do this? cd php4 cvs upd and a day later do this? cd php4/ext/standard cvs upd What is the checkout date here? I've no idea, do you? I think we just should not allow submissions with test results if they're not made by a snapshot or our phport.sh thingy for automatic testing. Sounds a lot more reliable :) Sander -- PHP Development Mailing List http://www.php.net/ To unsubscribe, visit: http://www.php.net/unsub.php
RE: [PHP-DEV] RfC: version names
Sascha: $ cd php4; cvs upd and a day later $ cd php4/ext/standard; cvs upd What is the checkout date here? It should be the time of the second update, if anything. Now it /is/ possible to find this out (by getting the most recent modification date of any CVS control file in the working directory), but that is probably too non-portable to be acceptable as part of the build system. (Okay, so you could write a mini C util to do it that relied only on POSIX, but that's still quite fiddly to get right.) However that's still going to be misleading, because any report against it has a reasonable chance of being bogus anyway, I'd have thought. It's difficult to detect, though. Derick's suggestion of only allowing submissions against at worst (in terms of moving targets) a snapshot build is probably a fair compromise? Cheers, James --- Outgoing mail is certified Virus Free. Checked by AVG anti-virus system (http://www.grisoft.com). Version: 6.0.449 / Virus Database: 251 - Release Date: 27/01/2003 This e-mail has been scanned for all viruses by Star Internet. The service is powered by MessageLabs. For more information on a proactive anti-virus service working around the clock, around the globe, visit: http://www.star.net.uk -- PHP Development Mailing List http://www.php.net/ To unsubscribe, visit: http://www.php.net/unsub.php
RE: [PHP-DEV] RfC: version names
I've no idea, do you? I think we just should not allow submissions with test results if they're not made by a snapshot or our phport.sh thingy for automatic testing. If that phport.sh thingy reliably ensures the integrity of the source files, it could work. Have CVS $Id$'s been discussed yet? This line will give you the timestamp from the latest checked in file: find . -type f|xargs grep '$Id: ' |grep -v Binary |\ sed 's#.*\([12].../../.. ..:..:..\).*#\1#'| sort | tail -1 - Sascha -- PHP Development Mailing List http://www.php.net/ To unsubscribe, visit: http://www.php.net/unsub.php
[PHP-DEV] RE: [PHP-QA] RE: [PHP-DEV] RfC: version names
This line will give you the timestamp from the latest checked in file: find . -type f|xargs grep '$Id: ' |grep -v Binary |\ sed 's#.*\([12].../../.. ..:..:..\).*#\1#'| sort | tail -1 This should also work with cvs export, does it? Can we add this to build system as Source Date so we have it in addition to the version string and the build date in phpinfo()? Regards, Kai -- PHP Development Mailing List http://www.php.net/ To unsubscribe, visit: http://www.php.net/unsub.php
Re: [PHP-DEV] RfC: version names
A very simple solution would simply be to do something like ls --full-time configure that allows you to find at what exact time buildconf was used so pretty much when it was built. This would be much easier to implement since you don't have to touch CVS at any time for that... right ? On Fri, 2003-01-31 at 08:29, Kai Schröder wrote: Hi all, I'm working on a new framework for analyzing results of make test for QA Team (see php-qa list). Talking with Sebastian Nohn, I realized, that we can't differ the versions in detail. That means, that we can't say which source packet was used to build php. Is there any way to name the versions 4.3.0 (release), 4.3.1-dev (taken from cvs, additional timestamp would be nice), 4.3.1-snapMMDDHHii (taken from snaps.php.net) and 4.3.1-win32-snapMMDDHHii (Win32 Snapshot)? This will make it more easy to follow the development and ignore allready fixed bugs. The goal is to find out which commit is the reason for the failure on which systems. If this works fine, we are able to add detailed comments to bugs.php.net and reopen automatic already closed bugs. Regards, Kai -- Xavier Spriet [EMAIL PROTECTED] -- PHP Development Mailing List http://www.php.net/ To unsubscribe, visit: http://www.php.net/unsub.php
RE: [PHP-DEV] RfC: version names
That can be done, but that means 12 commits a day for a single file. I dont think that's a good idea. Is there some way we can harness CVS keyword subsitutuion in a case like this? John -- PHP Development Mailing List http://www.php.net/ To unsubscribe, visit: http://www.php.net/unsub.php