Обсуждение: SQL standards in Mysql
Here's one Mysql developer's response to adding (fixing) the integer/bigint/tinyint types to their CAST function: http://bugs.mysql.com/bug.php?id=34562
Ken Johanson wrote: > Here's one Mysql developer's response to adding (fixing) the > integer/bigint/tinyint types to their CAST function: > > http://bugs.mysql.com/bug.php?id=34562 So they are anal too, but in the opposite direction? -- Alvaro Herrera http://www.CommandPrompt.com/ The PostgreSQL Company - Command Prompt, Inc.
On Fri, Feb 22, 2008 at 1:57 PM, Alvaro Herrera <alvherre@commandprompt.com> wrote: > Ken Johanson wrote: > > Here's one Mysql developer's response to adding (fixing) the > > integer/bigint/tinyint types to their CAST function: > > > > http://bugs.mysql.com/bug.php?id=34562 > > So they are anal too, but in the opposite direction? No, they're just hopelessly lost and making good time. Or lazy. Or both. I'm not sure. There's example after example of things in the mysql bug database that should make anyone considering it as a database engine cringe and walk away shaking their head. for instance, someone JUST submitted this bug: http://bugs.mysql.com/bug.php?id=34231 Which is about a problem with the mysql packages including a statically linked zlib which makes it impossible to compile php against it. Problem is, their inability to properly package / not package zlib with mysql is an ongoing problem. As far back as 2004 it showed up. Then was fixed, then showed up again, then was fixed. Here's the most recent "we fixed it! Oh shit it's broken again" bug record: http://bugs.mysql.com/bug.php?id=15255 Before that we had http://bugs.mysql.com/bug.php?id=3700 After that we had http://bugs.mysql.com/bug.php?id=28569 and now the latest reports. It's like the Keystone Kops create RPM packages reading those messages. Pick any subject area. Foreign keys, SQL compliance, and on and on and you'll find dozens of the same bugs coming and going as one person stomps it and another apparently reinstates it.
On 2008-02-22 16:13, Scott Marlowe wrote:
> There's example after example of things in the mysql bug database that should make anyone considering it as a
databaseengine cringe and walk away shaking their head.
I don't understand why anyone wanting a real SQL DB would pick MySQL.
Four years ago I knew nothing about SQL (I still pronounce it S-Q-L),
but wanted to set up a real DB with the maximum flexibility. So, I:
1. Asked a friend in the DB world (primarily Oracle) what he
recommended, and he said "MySQL".
2. I then bought a book on SQL ("Using SQL", by Rafe Colburn, ISBN
0-7897-1974-6, © 2000 Que Publishing), which mentions every common SQL
DB __except__ PostgreSQL.
So, I went with PostgreSQL. Why? From the book, it was clear that
MySQL lacked so many features of a decent SQL DB. In particular (at the
time) VIEWs and sub-selects. I didn't know much about VIEWs and
sub-selects, but it appeared to me that they were pretty
important/powerful features that any SQL DB should have. PostgreSQL was
the only one left standing (at least on my budget). (I ran it on a
Pentium 233 for a year before I upgraded the hardware.)
As of now, MySQL has VIEWs and sub-selects, but there appear to be a
number of other little "gotchas" that lurk (which the original poster of
this and the related threads has so amply illustrated). One of the
advantage of standards-compliant software is that, while you may be
surprised by some feature, that feature has been examined by a number of
people and (typically) found to be the best way of being consistent in a
broader view, rather than a feature that has been written (or not) for
the sake of expediency in a particular implementation. Further, many
standards are features that, while sometimes not met by existing
implementations, are at least a goal of consistency and functionality
that is aspired to (and usually planned for in future releases).
The sad fact is that we live in a world of expediency (not to mention a
bit of hype). Many people don't even do the elementary research that I
did before glibly picking a DB for their server, in their rush to be the
next dot-com (or other) success. As a result, PostgreSQL is not
supported by some software packages. For example, I think phpBB is the
only major message board software that supports PostgreSQL (see
http://www.phpbb.com/about/features/compare.php ), and in fact has for
some time. Of course, they have a DB abstraction layer (wow, what an
concept!), which allows them to easily support a number of DBs. Of
course, what does that tell you about the level of design and
professionalism of phpBB versus the others? Note that I have no
connection with the phpBB project, and in fact find many of the
developers arrogant. I'm just saying that any software that doesn't
support a number of DBs probably wasn't designed and/or implemented
properly.
Hopefully, some event in the near future will tip the scales in the
public perception of SQL DBs, and PostgreSQL will get better respect.
-- Dean
--
Mail to my list address MUST be sent via the mailing list.
All other mail to my list address will bounce.
On Fri, Feb 22, 2008 at 7:39 PM, Dean Gibson (DB Administrator) <postgresql@ultimeth.com> wrote: > So, I went with PostgreSQL. Why? From the book, it was clear that > MySQL lacked so many features of a decent SQL DB. In particular (at the > time) VIEWs and sub-selects. Note that unless someone's done some hacking recently, mysql uses nested loops for subselects. Which is fine for a few dozen or hundred entries. not so much for 1M rows. Handling subselects is as much about performance as it is about convenience. There was a time a few years ago when oracle would beat the pants off postgresql or mysql if you did this: delete from tablea where id in (select id from tableb); when tableb was pretty large. PostgreSQL hackers fixed that issue some time ago. MySQL, as far as I know, has no plans to fix their poor performance. Because somebody checked off subselect and that's all they needed, a check box. > the only one left standing (at least on my budget). (I ran it on a > Pentium 233 for a year before I upgraded the hardware.) Oh man, this takes me back. My first two postgresql servers were both pg v6.5.3 or so and ran on: Pentium-100 with 64 Meg of ram, Quad CPU Sparc-20 (4x50MHz CPUs) with a stack of SCSI hard drives running the last version of RH to support sparc back then. > supported by some software packages. For example, I think phpBB is the > only major message board software that supports PostgreSQL (see > http://www.phpbb.com/about/features/compare.php ), and in fact has for > some time. Of course, they have a DB abstraction layer (wow, what an > concept!), which allows them to easily support a number of DBs. Of > course, what does that tell you about the level of design and > professionalism of phpBB versus the others? Note that I have no > connection with the phpBB project, and in fact find many of the > developers arrogant. I'm just saying that any software that doesn't > support a number of DBs probably wasn't designed and/or implemented > properly. Well, somebody needs to teach them how this thing called patch works. The fact that all the hacks are installed by hand makes me queezy about messing with any modifications. Note that w-agora is a pretty nice bbs that supports mysql, oracle, and pgsql. I like the interface, but it doesn't look like there's a lot of recent development on it. I like the fact that it uses a nice threaded interface and can work with oracle, pgsql, or mysql. > Hopefully, some event in the near future will tip the scales in the > public perception of SQL DBs, and PostgreSQL will get better respect. I think it's slowly happening, not one big event, but a bunch of little experiences adding up. I think one of the things helping that in the long run is the number of very capable DBAs that are slowly coming to pgsql from other dbs, like DB2 or Oracle or Informix.
On Fri, Feb 22, 2008 at 11:03 PM, Mail Delivery Subsystem <mailer-daemon@googlemail.com> wrote: > This is an automatically generated Delivery Status Notification > > Delivery to the following recipient failed permanently: > > postgresql@ultimeth.com > To: "Dean Gibson (DB Administrator)" <postgresql@ultimeth.com> Look, if you're going to send email to this list from an address that doesn't accept email from anything but this list, and since this list by default is a "reply to all" list (i.e. people hit reply to all) the LEAST you can do is reconfigure your Mail client to change the reply to field to point to whatever list it is you are sending to. Or have the decency to program your MTA to just throw those messages away. Bouncing messages from a public list is kinda rude.
Ken Johanson <pg-user@kensystem.com> writes: > Here's one Mysql developer's response to adding (fixing) the > integer/bigint/tinyint types to their CAST function: > http://bugs.mysql.com/bug.php?id=34562 Hmm ... while I'm certainly not someone to defend mysql on a regular basis, I can see their point of view here. They are supporting a limited (and clearly documented) subset of the SQL spec in this area, so it's surely a feature request rather than a bug that they don't support more. The disagreement seems to come down to what the priority of the feature addition ought to be. We have a few sore spots of our own on questions like this one, so I'm disinclined to throw the first stone ... regards, tom lane
On Saturday 23 February 2008 07:50, Tom Lane wrote: >Hmm ... while ... > so I'm disinclined to throw the first > stone ... Meanwhile, Throw cones, not stones. http://cfx.kymi.com/lotsacones.jpg These things/projectiles hurt not so much. And it's fun ! BR, -- Aarni Ruuhimäki
On 2008-02-22 21:34, Scott Marlowe wrote:<br /><br /><blockquote cite="mid:dcc563d10802222134i5375ecafpaf33dcf329536d01@mail.gmail.com"type="cite"><pre wrap="">Bouncing messages from a publiclist is kinda rude. </pre></blockquote> No more so, than sending two copies of your reply to me, because you don'tgo up to your mailer's "To:" line and manually delete the extra address (as I do on EVERY reply I send to this list).<br/><br /> In fact, if you do that, you won't get the bounce. I do it out of consideration for others: since _I_would prefer to not receive two copies of replies, I assume that _others_ prefer the same. I don't whine about it; Ijust do it.<br /><br /> So, in return for that consideration, I get your reply. How did getting the bounce hurt you oryour computer, or make any extra work for you, more than I would have to do when I get a duplicate message????<br /><br/> Perhaps I've been working with computers too long. I've been PAID as a full-time software developer for the past40 years (45 if you count part-time employment in college), and I'm AMAZED at the amount of intolerance I see on theInternet with respect to eMails. Some people whine because the reply is at the top of the message rather than at thebottom. There are perfectly good reasons for replying at the top OR at the bottom, depending upon the circumstances. Other people whine because the sender does not wrap his/her eMail at 76 columns, or because the eMail is ALLCAPS, or some other imagined slight. Being an ADAPTABLE human being, rather than UPSET people who aren't similarly inclined,I simply ADAPT to the environment in most cases.<br /><br /> In this particular case:<br /><br /><blockquote cite="mid:dcc563d10802222134i5375ecafpaf33dcf329536d01@mail.gmail.com"type="cite"><pre wrap=""> Look, if you're going to send email to this list from an address that doesn't accept email from anything but this list, andsince this list by default is a "reply to all" list (i.e. people hit reply to all) the LEAST you can do is reconfigureyour Mail client to change the reply to field to point to whatever list it is you are sending to.</pre></blockquote>A good idea, but I use this eMail address to reply to multiple pgsql-xxx lists. That would requirea separate sender address for each list, but I can do that if I'm really hurting other people. It just means thatI will have to delete the duplicate replies, rather than have to hear about the problems (which have yet to be identified)the bounces causes other people. I'll consider it.<br /><br /><br /><blockquote cite="mid:dcc563d10802222134i5375ecafpaf33dcf329536d01@mail.gmail.com"type="cite"><pre wrap="">Or have the decency to programyour MTA to just throw those messages away. </pre></blockquote> That's a universally discredited idea among mail administrators: You either reject unwanted mail _during_the_SMTP_dialog_, or you forward it on to the recipient. The formeris strongly preferred for dealing with spam (which is why I bounce non-list replies), to avoid the bandwidth, storage,and other issues.<br /><br /> Now, for anyone who can't deal with the above, either:<br /><br /> 1. don't reply tomy posts;<br /> 2. edit the "To:" line in your replies to me; or<br /> 3. send me your eMail address and I will manuallywhitelist you (note that the whitelisting will disappear if some spammer uses your eMail address to circumvent thebounces). In fact, I'm surprised that spammers don't already do that using one of the administrator accounts here. MyeMail address here has been one of the largest targets for the spam my SMTP server rejects over the past few years, andI haven't even been very active here.<br /><br /> Sincerely, Dean<br /><pre class="moz-signature" cols="72">-- Mail to my list address MUST be sent via the mailing list. All other mail to my list address will bounce.</pre>
On Mon, Feb 25, 2008 at 9:08 PM, Dean Gibson (DB Administrator) <postgresql@ultimeth.com> wrote: > > On 2008-02-22 21:34, Scott Marlowe wrote: > > > Bouncing messages from a public list is kinda rude. > > No more so, than sending two copies of your reply to me, because you don't > go up to your mailer's "To:" line and manually delete the extra address (as > I do on EVERY reply I send to this list). It is considered polite to follow the customs of the groups in which one participates. For example, if you meet with Japanese people as part of your work you should bow whilst exchanging business cards, take care to read the card you are given, and not place it in the back pocket of your trousers. In this group we use the mj2 mailing list manager which by default will not send you a direct copy of any message which also has your email address explicitly listed. We prefer to use reply-all when responding to people, and that has become the custom here. It would be appreciated if you respected that custom during your time with us. Regards, Dave
Dean Gibson (DB Administrator) wrote: > On 2008-02-22 21:34, Scott Marlowe wrote: > >> Bouncing messages from a public list is kinda rude. >> > No more so, than sending two copies of your reply to me, because you > don't go up to your mailer's "To:" line and manually delete the extra > address (as I do on EVERY reply I send to this list). This is twice in as many days my old iLamp mail machine has been set aflame by the fires of a heated discussion about how a mailing list's reply-to is set. It gets hot enough just running Thunderbird. If you're going to continue this off-topic discussion, might I suggest taking it off-list? Interestingly, yesterday's flame-war took place because someone was adamant about just the opposite of your argument. Colin
On Mon, Feb 25, 2008 at 3:29 PM, Dave Page <dpage@postgresql.org> wrote: > On Mon, Feb 25, 2008 at 9:08 PM, Dean Gibson (DB Administrator) > <postgresql@ultimeth.com> wrote: > > > > On 2008-02-22 21:34, Scott Marlowe wrote: > > > > Bouncing messages from a public list is kinda rude. > > > > No more so, than sending two copies of your reply to me, because you don't > > go up to your mailer's "To:" line and manually delete the extra address (as > > I do on EVERY reply I send to this list). This is to Dean. I'm not editing my to: line on every single post. I'll hit reply to all. This list is designed to have it work that way, so that if the mailing list is running slow, the people participating in the discussion can keep up with it and if you're having problems now you don't have to wait 8 hours for the mailing list machine to get restarted or whatever it takes to make it behave sometimes. Also, every email client I've used in the last 5 years is smart enough to take the same message from two sources (i.e. the list / and the sender) and notice they're the same and NOT show them twice. It's not rocket science, and it's not uncommon, and it's not hard. If your client doesn't have that ability, I'd have to wonder what you're using. I think even pine can do this. It's just as easy to have your email program simply take anything that comes into it and toss it into the garbage and ignore it than to send a bounce message that uses up resources and bandwidth saying you don't want email from the people on the list, just the list. Especially since the way THIS list works, by generally accepted standards, is to hit reply to all. You'll notice I didn't answer you earlier. That's because you (Dean) have been put on my ignore list so I don't accidentally reply to you when you post, since it's the simplest and easiest way to stop getting bounce messages from you. Note this means I will miss your response, so don't bother sending one from your current email address to me, or to me through the list. You completely ignored my first post about this, and I really don't feel like having a complex involved conversation on basic email courtesy. Just one simple change in your software and voila it drops the dups instead of sending bounce messages. Good day.
On Mon, Feb 25, 2008 at 9:01 PM, Dean K. Gibson <random4@mailpen.com> wrote: > You are missing the point of why I do this. If it weren't for spammers No, I'm not. You're tilting at windmills. If you post to a public list, you will get spammed on that email address and there's not a lot you can do about that. What you can do is use basic spam detection and deletion software, like spam assassin, to sort the wheat from the chaff. I use gmail to read the lists. It's not perfect by any means. But I see none of the thousands of spam messages that hit my inbox everyday.We had spam assassin the last place I worked and itwas nearly 100% correct on getting spam, and it's fairly easy to set it up to be able to learn spam from you telling it any that slipped through. P.s. I've been using the internet since the mid 80s, and programming since the late 70s too. I remember a spam free internet, a spam free usenet even. Wish it could come back, but know that it won't.
On Mon, 25 Feb 2008 21:26:50 -0600 "Scott Marlowe" <scott.marlowe@gmail.com> wrote: > P.s. I've been using the internet since the mid 80s, and programming > since the late 70s too. I remember a spam free internet, a spam free > usenet even. Wish it could come back, but know that it won't. Damn that Canter and Siegel! -- D'Arcy J.M. Cain <darcy@druid.net> | Democracy is three wolves http://www.druid.net/darcy/ | and a sheep voting on +1 416 425 1212 (DoD#0082) (eNTP) | what's for dinner.
Dean Gibson (DB Administrator) wrote: > ...For example, I think phpBB is the only major message board software > that supports PostgreSQL (see > http://www.phpbb.com/about/features/compare.php ), and in fact has for > some time. Of course, they have a DB abstraction layer (wow, what an > concept!), which allows them to easily support a number of DBs. Of > course, what does that tell you about the level of design and > professionalism of phpBB versus the others?... In most cases it tells me is that the developers are willing to limit themselves to the lowest-common-denominator of features among the supported databases in exchange for the ability to run on a variety of backends. This is a perfectly legitimate decision for products where only basic database features are necessary. What I find scary is complex products that have been "ported" from something like mySQL to PostgreSQL. This often means that the product has severely limited its use of appropriate PostgreSQL features in order to remain compatible with the other backends. Cheers, Steve
Dean Gibson (DB Administrator) wrote: > On 2008-02-22 21:34, Scott Marlowe wrote: > > > Bouncing messages from a public list is kinda rude. > > > No more so, than sending two copies of your reply to me, because you > don't go up to your mailer's "To:" line and manually delete the extra > address (as I do on EVERY reply I send to this list). > > In fact, if you do that, you won't get the bounce. I do it out of > consideration for others: since _I_ would prefer to not receive two > copies of replies, I assume that _others_ prefer the same. I don't whine > about it; I just do it. > > So, in return for that consideration, I get your reply. How did getting > the bounce hurt you or your computer, or make any extra work for you, > more than I would have to do when I get a duplicate message???? I am surprised no one suggested majordomo's 'eliminatecc' option: http://mail.postgresql.org/mj/mj_wwwusr?user=&passw=&list=GLOBAL&func=help&extra=set Avoid courtesy copies The eliminatecc setting controls courtesy copy elimination. Ifthis setting is enabled, and your address appears in the To:orCc: headers of a posted message, Majordomo will not send anadditional copy to you. This helps to cut down on many ofthoseannoying duplicates that are often received, but it deprives youof the additional processing that Majordomo does ona message(subject prefixes, additional headers, etc.). eliminatecc - turns CC elimination on noeliminatecc -turns it off This does what the requestor wants, namely not send email from the list if they are already receiving the email as a reply. (It is spam removal or elimination of duplicates that is the problem?) > Perhaps I've been working with computers too long. I've been PAID as a > full-time software developer for the past 40 years (45 if you count > part-time employment in college), and I'm AMAZED at the amount of > intolerance I see on the Internet with respect to eMails. Some people > whine because the reply is at the top of the message rather than at the > bottom. There are perfectly good reasons for replying at the top OR at > the bottom, depending upon the circumstances. Other people whine > because the sender does not wrap his/her eMail at 76 columns, or because > the eMail is ALL CAPS, or some other imagined slight. Being an > ADAPTABLE human being, rather than UPSET people who aren't similarly > inclined, I simply ADAPT to the environment in most cases. I think a lot of the finickiness comes from the fact that emails often go to thousands of people, which does require more work from the email author. It is the cost of being able to communicate with that many people at once. > Mail to my list address MUST be sent via the mailing list. > All other mail to my list address will bounce. Let me add that just trashing all email from you is an option many might choose. You seem more concerned with making things easy for yourself and not adequately considering the thousands of people who are you communicating with, and replying to your emails trying to help you. -- Bruce Momjian <bruce@momjian.us> http://momjian.us EnterpriseDB http://postgres.enterprisedb.com + If your life is a hard drive, Christ can be your backup. +