留言板

Is Liferay 7 GA3 ready ?

thumbnail
Eric COQUELIN,修改在7 年前。

Is Liferay 7 GA3 ready ?

Expert 帖子: 254 加入日期: 13-11-3 最近的帖子
Dear all,

After testing Liferay 7 GA3 for a while, I have the feeling that this version is absolutely not ready for production obsiouly but not even ready for development. I can see on the forum people asking for different things but no one really complaining (at least I didn't notice), then I'm wondering if my feeling is correct or if I'm just complaining (like French use to do emoticon )

Let me explain you what's wrong (in my point of view):
  • Starting time is now much longer than Liferay 6.2
  • Coming with OSGI, each time you update your code, bundle is updated and restarted. Now, deployment time takes longer, development work too
  • Theming recommends using the new generator, each modification performed on a CSS requires now recompilation and deployment. About 15 seconds to get each modification available on the portal
  • It looks like Liferay developers forgot about Windows users and most of the tools are not optimized or even block (I suggest you try doing a theme on Windows and after 'gulp watch', kill the task. The server keeps a lock on files and you can't restart 'gulp watch' unless you restart your server or change your theme from the portal)
  • Have you tried modifying a web content using structures, some fields are not modifiable (but should be) and the "save" bar remains visible on the top of the modal without sticking to bottom line
  • Don't try migration when you have web content/structure with images in Liferay 6.2. Your images are migrated but the web content can't find them any more
  • Migration from Liferay 6.2 to Liferay 7 GA3 for a very simple/light web site (just content) took me 40 minutes to complete
  • Page loading is sligthly longer than with 6.2 (even after cache updates)
  • Development using Eclipse requires more RAM than before. Now you have a new process using up to 500Mb running in addition to your normal Eclipse and server process and it seems to be used for OSGI development. Talking to Liferay staff, they recommend using 'blade deploy -w' instead to save resources
  • Documentation is not aligned with what is available in GA3, I hope it is the case for DXP. As a result, you spend hours to understand why your code is not running as expected...
  • Documentation is richer than before BUT if you're a newbie, good luck! I wish Liferay staff could take some examples from well written documentation such as Angular or Django
  • Javascript ES6: nice feature but each compilation requires more than 10 seconds to complete. Unless you have time...


I'm a real fan of Liferay, I have been working for the last 3 years with 6.2. Some of you may think that I'm too much used to that version and refrain using the new one but before writing this ticket, I have been testing LR 7 during weeks. And I'm pretty sure I forgot some points... If you think I'm wrong, don't hesitate to reply as I'm looking forward explanations.

There are very nice improvements coming with Liferay 7 and I wish I could push it for my own development or even to my customers. I assume (hope) DXP is much more stable or even faster than CE but I feel like CE is probably the first version new customers test before asking for EE details. That's why I think Liferay should consider improving very quickly GA3.
thumbnail
Andrew Jardine,修改在7 年前。

RE: Is Liferay 7 GA3 ready ?

Liferay Legend 帖子: 2416 加入日期: 10-12-22 最近的帖子
Hi Eric,

I am a huge fan of Liferay as well. I have my frustrations with the product as you do but at the end of the day, I always find it to be better than the alternatives. And even with the work arounds and such that are required, I find it faster to build on that anything else (that I have used). I have some comments on your points -- some of it might help you.

1.Starting time is now much longer than Liferay 6.2

It doesn't have to be. You can disable all the modules that you don't need for your solution. For example, no blogs? or wikis? disable those bundles in OSGI and it will eliminate their activation cycles during startup. I helped someone on IRC with this problem and he cut his startup time by 2/3's!

2. Coming with OSGI, each time you update your code, bundle is updated and restarted. Now, deployment time takes longer, development work too

Personally I haven't found it to be any longer than it was with 6.2. The only reason I MIGHT agree with this is because in the older versions of Liferay I was able to use JRebel which would automatically reload the classes for me. It might be possible to do that with the OSGI container as well, but I haven't tried to figure it out yet. At any rate, I think that this point is arguable since even things like how crappy your machine is, or which OS you use (you mentioned windows -- *BARF*) can affect performance. My laptop is 5 years old running Linux and it's faster than most people new windows machines.

3. Theming recommends using the new generator, each modification performed on a CSS requires now recompilation and deployment. About 15 seconds to get each modification available on the portal

I haven't experienced this. In 6.2 and older I used emilio bergs instant theme deploy -- a tremendous time saver. In LR 7 the gulp watch is now part of the tool kit, so that should allow you to save and reload without compilation. Maybe a configuration error somewhere? In fact I went to NAS2016 in september and there was a tool that was demonstrated that basically allows you to develop and configure your theme outside the IDE .. allowing you to build your theme without deploy. I can't recall where it was exactly, but if you are interested I'll see if I can dig it up in my notes.

4. It looks like Liferay developers forgot about Windows users and most of the tools are not optimized or even block (I suggest you try doing a theme on Windows and after 'gulp watch', kill the task. The server keeps a lock on files and you can't restart 'gulp watch' unless you restart your server or change your theme from the portal)

No comment on this. I forgot about windows about 10 years ago now -- best memory I ever lost emoticon

5. Have you tried modifying a web content using structures, some fields are not modifiable (but should be) and the "save" bar remains visible on the top of the modal without sticking to bottom line

6. Don't try migration when you have web content/structure with images in Liferay 6.2. Your images are migrated but the web content can't find them any more

I would chalk 5 and 6 up to bugs. No one likes bugs, but they exist in software, ALL SOFTWARE. Software that appears like it is bug free is simply software with bugs that have not been found yet. I know that sounds like an excuse but with a product as large and feature rich as Liferay, you're going to have bugs. Couple that with such a major architectural shift in the 7.0 version and you have to expect there to be some pain.

7. Migration from Liferay 6.2 to Liferay 7 GA3 for a very simple/light web site (just content) took me 40 minutes to complete

Did you ever try migrating a theme from 6.1 to 6.2? It was a colossal pain in the ass. Mostly because the UI layer was so different between the two. The changes in 7 are even bigger so, yep, migration sucks. It always sucks. Brett Swaim did a talk on it at NAS and he said the same thing you did. The first couple of plugins took a long time to sort out. But once he had the pattern down he found that it was much quicker to migrate them. Don't forget that part of the issue a lot of us face with this new release is the learning curve. It basically resets Liferay Power Users to an Academic level of accomplishment. We all UNDERSTAND how to build applications in Liferay, but we need to learn again the practical skills to accomplish it.

8. Page loading is slightly longer than with 6.2 (even after cache updates)

Personally I have never found the default Liferay cache settings useable. In fact, the default caching (in JVM) implementation is know to suck balls. If caching is a big necessity for your application, I would consider offloading to something like Terracotta and then making sure to use something like VisualVM to monitor the cache statistics and tune accordingly. Again though, don't forget to TURN OFF ALL THE THINGS YOU DON"T NEED! This is a powerful new feature, something that was never available (without pain) in the past.

9. Development using Eclipse requires more RAM than before. Now you have a new process using up to 500Mb running in addition to your normal Eclipse and server process and it seems to be used for OSGI development. Talking to Liferay staff, they recommend using 'blade deploy -w' instead to save resources

Eclipse sucks. It's always been a memory pig -- and a CPU pig for that matter. Switch to intellij! Embrace the console! emoticon

10. Documentation is not aligned with what is available in GA3, I hope it is the case for DXP. As a result, you spend hours to understand why your code is not running as expected...

11. Documentation is richer than before BUT if you're a newbie, good luck! I wish Liferay staff could take some examples from well written documentation such as Angular or Django

Documentation for me has always been something lacking. The nice thing about the older versions is that a lot of people OUTSIDE of Liferay wrote far more useful posts than a lot of the stuff I found on Liferay's site (sorry Liferay). My frustration has always been that the docs I do find are too simplistic in their example and as you say, don't explain well. I chalk it up to Liferay is a software company, not a training company. Docs are important, but less important than fixes and features. I think the lack of examples for 7 goes back to what I said earlier. Liferay Power Users or Experts no long have the "hands on experience" to write about all the theoretical solutions. It'll take time to build up that knowledge base and pool of examples again. It sucks -- yes, I agree. But on the flip side, it force you to roll up your sleeves and dig deep into the source which will help you understand how the portal works internally much better than documentation ever will.

12. Javascript ES6: nice feature but each compilation requires more than 10 seconds to complete. Unless you have time...

No comment on this -- other than you gotta go with what industry is doing if you don't want to be left behind. I have not real experience with this -- do you think the compilation is something that is slow because of Liferay's implementation? or do you think it is an external dependency (library or something perhaps) that causes the pain?

.. I would echo a lot of your concerns though. GA3 has been out for a while now so I would expect a GA4 to be coming out shortly -- maybe someone from Liferay could comment? But if you ever worked with GA1, you'd think that GA3 was a god send emoticon. I think your approach is the right one and I am doing something similar. I am letting my clients know about the differences between 6.2 and 7 and letting them decide. One important caveat though is that 6.2 is no longer receiving updates. So anything you find in there that is not working will be on you to fix. There are bugs in 7, sure, but at least we can still report them and hope for fixes.

Just my two cents.
thumbnail
Eric COQUELIN,修改在7 年前。

RE: Is Liferay 7 GA3 ready ?

Expert 帖子: 254 加入日期: 13-11-3 最近的帖子
Thank you Adrew for taking time replying to my post.

It doesn't have to be. You can disable all the modules that you don't need for your solution. For example, no blogs? or wikis? disable those bundles in OSGI and it will eliminate their activation cycles during startup. I helped someone on IRC with this problem and he cut his startup time by 2/3's!


I'll try that and will revert back to you. Which IRC channel & server should I connect to (if possible)?

Personally I haven't found it to be any longer than it was with 6.2


There are two types of modifications that can be performed on a portlet in Liferay 6.2. Working with the MVC portlet, JSP modifications do not require redeployment and are very fast to be propagated. On the other hand, class modification takes longer, maybe as long as in LR 7. However, in LR7, any modifications, even on JSP, requires redeployment. At the end of the day, as you spent more time modifying the view, you spend more time on development than before for the same type of job

n LR 7 the gulp watch is now part of the tool kit


I use it but each SCSS modification requires compilation and it takes 15 seconds...

there was a tool that was demonstrated that basically allows you to develop and configure your theme outside the IDE


That sounds good but I never read anything about it emoticon

I forgot about windows about 10 years ago now -- best memory I ever lost


But from a business perspective, you can't ignore it

Did you ever try migrating a theme from 6.1 to 6.2?


There is a misunderstanding. I'm just talking about migration of database using the java command in tools/. I'm not even talking about theme or portlet migration. And in my case, theme is written without SDK and as a result, I have to rebuild everything. However, this point is not a big issue, migration is just a one shot operation, I was just surprised it took so long for a so small database

Eclipse sucks. It's always been a memory pig


Did I miss something? As far as I remember this is the official tool of Liferay. and unless I'm wrong, Developer studio is based on Eclipse. No?

Documentation for me has always been something lacking.


I'm happy you share my point of view. However, I do not agree with the priority. When it comes to pushing Liferay to a new customer, he will look at the documentation before subscribing because he wants to avoid long learning curve. Liferay should be easier to understand, at least for "simple" to "middle complex" applications. As a result, Liferay will be easier to sell.

No comment on this -- other than you gotta go with what industry is doing if you don't want to be left behind


I'm not saying ES6 sucks, not at all ! I'm just saying that I wish I could use it but modules using ES6 takes even longer to deploy. I'm already complaining about the time it takes for "normal" modules but if you add ES6 JS, it is even worse...

More generally, most of my issues may come from using Windows: there are many command line taks and we all know Windows is not as good as Linux for this kind of thing. I like Linux but the fact is that I have Windows so I have to deal with it. I know also some prospective customers who are using Windows for their development, how will they deal? Are we going to tell them that after purchasing Liferay, they should switch to Mac OS or Linux for productive development?

Hope Liferay will provide insights about all these things.
thumbnail
Andrew Jardine,修改在7 年前。

RE: Is Liferay 7 GA3 ready ?

Liferay Legend 帖子: 2416 加入日期: 10-12-22 最近的帖子
I'll try that and will revert back to you. Which IRC channel & server should I connect to (if possible)?


I am on freenode and the channel is #liferay. It's pretty quiet.. most people don't say anything, but I try to keep an eye on it in case someone comes along.

That sounds good but I never read anything about it


I haven't seen any references to it outside of the conference. I'll have a look through my notes and try to get back to you on what it was called and where you can get it from.

Re: WIndows -- But from a business perspective, you can't ignore it


I've ignored it quite successfully for about 7 years and I run my own business. I have a windows VM that I keep for Office products (LibreOffice and OpenOffice just don't cut it) but everything else I am able to do without it. I often find myself working with clients where the entire dev team is using Windows and I still use Linux. Java man -- it's platform independent! When I first started with Linux I was afraid to leave windows so I just ran a VirtualBox VM with ubuntu and every day would work in the VM. Eventually I just switched from Windows as a host to Linux as a host. You can be part of an organization that standardizes on Windows and still use Ubuntu (as I did) for your development work. In fact, I find that Windows itself (for Windows 7 at least) runs BETTER in a VM than it does as the host OS

There is a misunderstanding. I'm just talking about migration of database using the java command in tools/. I'm not even talking about theme or portlet migration. And in my case, theme is written without SDK and as a result, I have to rebuild everything. However, this point is not a big issue, migration is just a one shot operation, I was just surprised it took so long for a so small database


Fair enough -- but it's usually a one time thing and a good excuse to go for a coffee emoticon Why it takes so long? No idea. What Database are you using? MySQL? I would try increasing the log levels for the upgrade and verify processes to see which steps are taking the longest to complete. Then you could have a look at the class to see what it is doing. It might provide some insight -- or allow you to figure out where the bottlenecks are and how to remove them

Did I miss something? As far as I remember this is the official tool of Liferay. and unless I'm wrong, Developer studio is based on Eclipse. No?


it is -- but honestly every engineer I have spoke to at Liferay uses Mac/Linux and intellij or sublime. The IDE is very useful for having all the little GUIs for the configuration files but I am willing to bet after 3 years with Liferay you don't really need those tools anymore. Intellij is infinitely faster than eclipse (and more stable) -- I have found my pace of delivery increase since I switched to it, even with the loss of the Liferay Studio plugins.

There are two types of modifications that can be performed on a portlet in Liferay 6.2. Working with the MVC portlet, JSP modifications do not require redeployment and are very fast to be propagated. On the other hand, class modification takes longer, maybe as long as in LR 7. However, in LR7, any modifications, even on JSP, requires redeployment. At the end of the day, as you spent more time modifying the view, you spend more time on development than before for the same type of job


Missed this one so it's at the bottom emoticon. Ah, you are referring to the auto-deploy. That, I believe only actually works with Tomcat as your app server. JRebel provides the same feature but also auto-deploy for changes to the Java classes . Your point is fair -- though I still say that you might have a gain by using JRebel, I just have not tried yet. It's all more cost -- but honestly, the cost is negligible compared to productivity gains.

There are a lot of gains with OSGI too that you need to remember. First and foremost the stability that the OSGI container gives you when dependencies "go missing". There is also the fact that you no longer have to restart your server when you want to change the registered service -- or the fact that you can apply logic to have multiple services registered and set the priority rather than a single registered service in a static spring mapping.

It's a brave new world. There are many kinks to still be worked out. Personally I am more upset about the absence of clustering and audience targeting than I am the extra time it takes to compile and deploy.
thumbnail
Eric COQUELIN,修改在7 年前。

RE: Is Liferay 7 GA3 ready ?

Expert 帖子: 254 加入日期: 13-11-3 最近的帖子
it is -- but honestly every engineer I have spoke to at Liferay uses Mac/Linux and intellij or sublime.


This is exactly where the problem comes from. If no tests or/and no one use Windows for day to day development/test, how can they realise there is a BIG issue? Fair enough, if development under Windows is counter productive, just let people know. Or make sure, it is as productive as it is on other platform. Right now, documentation doesn't state anything about it.

This morning, I have set up Virtual Box on my Windows computer and have deployed Ubuntu 16. After installing all the stuff, I tried theme building using generator and the watch function. Linux is about twice faster than Windows. Looking at the different taks, it looks like the SASS compiler is the main responsible for low performance (I'm using gem sass).

I have not yet tested the jar/OSGi module deployment but I'm pretty sure I'll get better deployment time than with Windows.

As a result, at least with GA3 and current tools, I think developers should only consider Linux (or like) to perform Liferay 7 development.
thumbnail
David H Nebinger,修改在7 年前。

RE: Is Liferay 7 GA3 ready ?

Liferay Legend 帖子: 14914 加入日期: 06-9-2 最近的帖子
Eric COQUELIN:
it is -- but honestly every engineer I have spoke to at Liferay uses Mac/Linux and intellij or sublime.


This is exactly where the problem comes from. If no tests or/and no one use Windows for day to day development/test, how can they realise there is a BIG issue?


There is a difference between testing tools to ensure they work and using them for day to day development tasks. Greg and his team do a lot of work on the IDE and Developer Studio (both Eclipse based) and they care a great deal about what problems the developers have with them.

That said, they have no control over the performance of gem sass on windows or whether the right version of gulp is installed for the "gulp watch" command to work.

Just like Liferay is application server neutral, taking the position that a company can use the app server they want and not use one mandated by Liferay, the development tools are the same way. We don't tell folks they have to use the IDE or Dev Studio; if you don't have a tool we're happy to give you one with additional tools to make developing for Liferay easier, but it by no means is meant to be a message like "only use our tools". If you have a tool that you like better, that works better in your environment, you're welcome to use it, we're not going to get in the way.

This morning, I have set up Virtual Box on my Windows computer and have deployed Ubuntu 16. After installing all the stuff, I tried theme building using generator and the watch function. Linux is about twice faster than Windows. Looking at the different taks, it looks like the SASS compiler is the main responsible for low performance (I'm using gem sass).


This kind of points out the problem that Liferay often faces. Issues you're seeing are not Liferay's fault, they're environmental. The tools, etc. work, they're just hamstrung by the operating system itself. I know on dev.liferay.com the discussion about the SASS compiler has a special section for windows because the tools themselves, and not Liferay, perform poorly on Windows.

Frankly no one ever asks what the best environment is for development. Honestly I don't believe organizations would invest money in handing out macbooks if Liferay said the tools work best there; developers are often stuck with the boxes their IT overlords give them when they join the company and are expected to use them to get the work done. Liferay ensures the tools work on the platforms so developers can be successful, but they are not accountable for the environments the developers get to use.
thumbnail
David H Nebinger,修改在7 年前。

RE: Is Liferay 7 GA3 ready ?

Liferay Legend 帖子: 14914 加入日期: 06-9-2 最近的帖子
Hey, Eric, thought I'd throw my $0.02 in...

Eric COQUELIN:
1. Starting time is now much longer than Liferay 6.2


With each release, this continues to get better. If you were a DXP customer, you would be up to Service Pack 1 and Fix Pack 9, all of these w/ important bug fixes and performance enhancements.

I'm not sure when the next GA is coming out, but I'm sure a lot of these will be there.

2. Coming with OSGI, each time you update your code, bundle is updated and restarted. Now, deployment time takes longer, development work too


The same was true if you changed java code in a portlet under 6.x; you still had to deploy/sync it and wait for the app server to process the change. And just as in regular portlet development, once you get your skis under you on OSGi development the number of intermediate deployments should go down.

3. Theming recommends using the new generator, each modification performed on a CSS requires now recompilation and deployment. About 15 seconds to get each modification available on the portal


This actually was a request from front-end developers. They typically don't use maven, ant, and all of the other tools we as back-end developers are comfortable with. Many times they're using simply Sublime for editing and the rest of the toolset (gulp, npm, etc). To that end they leverage the gulp watch command which reflects changes in theme files in the portal almost instantaneously.

4. It looks like Liferay developers forgot about Windows users and most of the tools are not optimized or even block (I suggest you try doing a theme on Windows and after 'gulp watch', kill the task. The server keeps a lock on files and you can't restart 'gulp watch' unless you restart your server or change your theme from the portal)


If anything that is a cultural thing based off of the tools, not off of a specific decision of Liferay's. For example, they don't author or maintain the gulp tool, that belongs to another group. Same with many of the other tools. I'd recommend that you verify that you have the latest and greatest of all of the available tools and don't just rely on the installed version that you might have started with.

5. Have you tried modifying a web content using structures, some fields are not modifiable (but should be) and the "save" bar remains visible on the top of the modal without sticking to bottom line


Not sure what issue you're having here. I don't work with these as often as I should, but I haven't heard about this problem before.

6. Don't try migration when you have web content/structure with images in Liferay 6.2. Your images are migrated but the web content can't find them any more


This is another issue that I haven't heard of yet, but I guess there may be something in issues.liferay.com on it. I can say that many of the upgrades that I have been through so far, many of the failures were related to, well, lets just say challenges in the original data; folks doing some 'edge' things in content or leveraging unpublished APIs or using known deprecated features to get things to work they way they want, well much of that just won't work so well during an upgrade.

7. Migration from Liferay 6.2 to Liferay 7 GA3 for a very simple/light web site (just content) took me 40 minutes to complete


So first, each upgrade comes with database changes. It's one of those reasons why we say folks should stay out of the database because it's subject to change, but that's a different discussion for another day.

Because of the database changes necessary to support all of the new functionality in Liferay 7 and in an effort to ensure the legacy data will still be available after the upgrade, there's some lengthy data manipulation steps necessary to get the legacy data into DXP.

40 minutes is really nothing to complain about; one client I'm working with takes a week for their data to go through the upgrade process simply because of the size of their database. It just is what it is, but fortunately it's a one-time thing that can be forgotten after you're done with it.

8. Page loading is sligthly longer than with 6.2 (even after cache updates)


This is along the lines of the performance improvements coming out regularly in DXP. I'm sure they'll get out in a future GA release, but if you can't wait DXP is an option.

9. Development using Eclipse requires more RAM than before. Now you have a new process using up to 500Mb running in addition to your normal Eclipse and server process and it seems to be used for OSGI development. Talking to Liferay staff, they recommend using 'blade deploy -w' instead to save resources


I really don't use Eclipse at all myself. I'm 100% Intellij now, it seems to be much better with the memory management than Eclipse.

10. Documentation is not aligned with what is available in GA3, I hope it is the case for DXP. As a result, you spend hours to understand why your code is not running as expected...


The documentation on dev.liferay.com targets both CE and DXP. If you're finding stuff that isn't right, you can either open tickets on it or, better yet, submit some documentation PRs for the doc team to review. I think part of the problem is that the guys just have a lot of work going on now working on documentation, the new certification exams, etc. Nothing is purposefully misleading, if anything I'd say it may just be an oversight.

11. Documentation is richer than before BUT if you're a newbie, good luck! I wish Liferay staff could take some examples from well written documentation such as Angular or Django


Documentation has long been a complaint, but as you've seen there is a lot more timely DXP doco available than what was there for previous releases. It goes to the huge amount of work the doco team has been pushing through to get them done.

It's definitely hard for newbies because now you're not just learning Liferay but often this is probably the first time you've heard about OSGi.

Personally I think as a community we shouldn't sit back and just wait for Liferay to generate doco, I feel that it's important that we in the community give back. For example, I put a 6 part blog series up on Liferay 7 development because I thought it could help fill some holes and help folks get a vision on how overall the OSGi development process works.

I don't expect I'll be alone in this effort, if anything I like to hope that folks are still getting up to speed and the community-generated info will start pouring out.

12. Javascript ES6: nice feature but each compilation requires more than 10 seconds to complete. Unless you have time...


Well, this is only an issue for the current browsers that do not yet have native support for ES6 yet. Chrome and FireFox (I believe) have engines in the pipeline that do not require transpilers on the JS. The other browsers will soon be in line and transpilation will go away.

I'm a real fan of Liferay, I have been working for the last 3 years with 6.2. Some of you may think that I'm too much used to that version and refrain using the new one but before writing this ticket, I have been testing LR 7 during weeks. And I'm pretty sure I forgot some points... If you think I'm wrong, don't hesitate to reply as I'm looking forward explanations.


We love the fans! emoticon

For all of the problems you find, please make sure you open a ticket on issues.liferay.com. If the problems don't get reported and/or voted on, Liferay won't really know that the problems are out there.

There are very nice improvements coming with Liferay 7 and I wish I could push it for my own development or even to my customers. I assume (hope) DXP is much more stable or even faster than CE but I feel like CE is probably the first version new customers test before asking for EE details. That's why I think Liferay should consider improving very quickly GA3.


If you're judging DXP on what you see in CE, well I'd encourage you to rethink that. CE and DXP have the same baseline code, but where there are fewer releases of CE, there have been lots of updates for DXP. Fixes and tuning efforts are still going on for Liferay and these improvements ship regularly in DXP.

From the perspective of evaluating DXP by looking at CE, I would say that you can review CE to get a feel for what you'll get from DXP, but DXP is so much more than EE releases in the past that were just CE + additional enterprise components. That's the reason why it is now Liferay DXP and not Liferay 7 EE, there is just so much more to it than what the product was in the past.
thumbnail
Eric COQUELIN,修改在7 年前。

RE: Is Liferay 7 GA3 ready ?

Expert 帖子: 254 加入日期: 13-11-3 最近的帖子
Hi David,

Thank you for replying. I think I already provided more details in the previous responses with Andrew.

Except one point...

That's the reason why it is now Liferay DXP and not Liferay 7 EE, there is just so much more to it than what the product was in the past.


I understand that from a marketing/business perspective, that was important for Liferay to push more in DXP than before so that the gap with CE gets bigger and bigger and users may prefer DXP than CE. But beware the gap gets not too large otherwise users will switch to another solution because not everybody can afford Liferay DXP. Community is a wealth for Liferay, we must not neglect it

I'm obviously looking forward GA4.
thumbnail
David H Nebinger,修改在7 年前。

RE: Is Liferay 7 GA3 ready ?

Liferay Legend 帖子: 14914 加入日期: 06-9-2 最近的帖子
Eric COQUELIN:
But beware the gap gets not too large otherwise users will switch to another solution because not everybody can afford Liferay DXP.


So I kind of disagree with this statement. Liferay DXP does have a price tag that comes with it, but with that price you get support, regular updates, priority for bug resolution, etc. That alone can actually make DXP cheaper than self support and maintenance with CE.

For non profits and charitable organizations, Liferay has special pricing.

And honestly, if you haven't talked to Liferay sales you may have an impression that Liferay is expensive that may not be completely accurate...

Community is a wealth for Liferay, we must not neglect it


Of course I agree with you here, I've been a community member for like 10 years now emoticon
thumbnail
Andrew Jardine,修改在7 年前。

RE: Is Liferay 7 GA3 ready ?

Liferay Legend 帖子: 2416 加入日期: 10-12-22 最近的帖子
Hey David,

Though you and I are pretty well ALWAYS on the same page, I do think Eric has made some good points. I'm really excited about what Liferay has to offer -- in all their versions -- but do feel that with this new direction it's a little less "open source" than it has been in the past. Personally, I would like to see Liferay continue with all the great open source-ness it has always done and maybe limit the cost and effort that it takes on as an organization to maintain that opensource-ness by leaning on the community more to support the community version. I think that the community benefits from this, but also Liferay might as well. Who's to say that a community member wouldn't contribute a new feature to -- say audience targeting -- that couldn't be rolled into the DXP stream and be made available to the enterprise clients. A lot of open source projects benefit from community contributions. So, from where I sit, I would like to see Liferay bring back community contributions for those projects but basically wash their hands of the support on them - like a donation of sorts.

I still urge all my clients to go with a licensed product for all the benefits you list above, but there are cases where the funds might be there, but they aren't willing to take the risk up front. Even better, independents like myself, who would love to -- say learn more about a feature to use it as a carrot to get a client to use Liferay, only to find that it is not available to me. So now I find myself saying to clients -- "yep, audience targeting is what you need. I'm pretty sure you can do x or y" -- pretty sure -- rather than "I've done this and this with it and I know that with a few tricks you can also do that.. etc"

Liferay has always been so good at making my job easier -- why start making it harder now! emoticon
thumbnail
David D Pierovich,修改在7 年前。

RE: Is Liferay 7 GA3 ready ?

Junior Member 帖子: 26 加入日期: 09-6-30 最近的帖子
I've been using DXP for a couple months now and am really impressed (not favorably) by the number of bugs I continue to encounter. It's become something of a running joke as I've compiled about 4 pages of bugs so far. I stopped recording everything Im finding, just too much to bother.

Hearing people question whether CE 7 GA3 is ready only begs the question for enterprise customers as well. I don't think DXP or its tools are really ready for prime time. I've been working with Liferay a long time and can't recall such a bug-fest in any previous version. Sure, I've had to submit tickets in the past, but nothing like I'm doing for DXP. I've submitted more bugs in two months of DXP than I've submitted in all previous versions combined, and I'm not submitting everything I find. Nobody needs that much brain damage and my role isn't to augment DXPs testing, it's to deliver solutions using the platform and tools.

I get that Liferay is in the middle of a mammoth transformation on its platform and tools, but the product needs to be rock solid if you're going to charge for it. Releasing a bug filled commercial product that is clearly premature doesn't instill confidence.
thumbnail
David H Nebinger,修改在7 年前。

RE: Is Liferay 7 GA3 ready ?

Liferay Legend 帖子: 14914 加入日期: 06-9-2 最近的帖子
Have all of the bugs been reported?

We're already up to fix pack 10 and 11 is just around the corner. SP1 is out and SP2 is also coming soon.

If the LESA tickets are opened on your issues, they'll get solved.
thumbnail
Daniele Baggio,修改在7 年前。

RE: Is Liferay 7 GA3 ready ?

Expert 帖子: 336 加入日期: 08-12-5 最近的帖子
mmmmm.. sounds shitty but true...
thumbnail
David H Nebinger,修改在7 年前。

RE: Is Liferay 7 GA3 ready ?

Liferay Legend 帖子: 14914 加入日期: 06-9-2 最近的帖子
The reality is that although Liferay has a big test suite, everyone ends up using Liferay in their own way and often issues in how they use it only come out when they do the testing.

When you report it, Liferay will get around to fixing the issue. If you don't report it, Liferay won't have an idea there's an issue without the report. The test cases will continue to pass and they'll believe all is good.

When the first milestones came out I encouraged everyone to do testing on how they use Liferay. Because this is a long standing issue, actually. If all you do is use the OOTB features w/ no changes, you're typically pretty good. But when you apply a customization, well then you're using Liferay differently than what they test for and those things may not be covered in a test case.
thumbnail
Andrew Jardine,修改在7 年前。

RE: Is Liferay 7 GA3 ready ?

Liferay Legend 帖子: 2416 加入日期: 10-12-22 最近的帖子
There is nothing I love more than people who hold Liferay to a higher standard than other software vendors. In fact, I love that they hold Liferay to a higher standard than themselves even. Let me guess, you'll say, "yeah but I don't sell software, it's different." -- really? you go to work every day as your charitable donation to the world of software? no one pays you to work? you've never cut a corner because of a deadline? you've never released a version of your work that wasn't flawless?

It's easy to pass judgment on other peoples work. 4 pages of bugs? you think that is a lot? from a product that is basically rearchitected from the ground up? From a product that has upwards of 700 features? 4 pages, hell 15 pages of bugs would be an accomplishment in my opinion. I'd bet moeny that teams you have worked on, with projects that are tiny in comparison to the size of Liferay, with features that are dwarfed by what Liferay offers have as many pages of bugs. From my experience (of which I have quite a bit spanning not just Liferay but many of its competitors) and working with clients, 90% of the time the "bug" they found is actually the user or implementor doing something wrong. I remember one conversation I had on day 1 with a new client a couple of years ago. He asked me about something and I explained that you use the serviceLocator in the template blah blah blah. He told me that it didn't work, so for his solution he created a custom portlet, that used velocity templating to render the result and then loaded his content from property files. I was shocked. He basically duplicated, in a worse way (you had to restart to update content) the Web Content Portlet. His claim that the feature he wanted didn't work was in fact completely false and there were several things that he was actually doing wrong. But by the time I got there, this feature, and several others had been black listed as the result of them "not working" or "full of bugs" when in point of fact it was a PEBCAK (Problem Exists Between Keyboard and Chair).

Liferay is not perfect and the bugs I come across frustrate me as well. I'm sure my clients get frustrated when they come across features that they have paid me to build that don't work perfectly with release 1. Welcome to software, you report it, and it gets fixed. The nice thing about Liferay as compared to many other vendors (at least from those I have worked with) is that if you really need to, you can fix it yourself. Even with the EE license you have access to the source code of the core product. You think Liferay's top competitors will give you that? And David's point is in fact case an point. When you license the product you basically buy a support contract. So if you have a problem, you report it, it gets fixed for you right away. Why you would then come to the forums to complain about the very thing you pay for is a little confusing.

If your rationale for a product being ready for primetime is based on whether or not it has bugs then we might as well all close our laptops and go home. There is no such thing as software without bugs, The best you can offer is software with bugs that have not been found yet. I still maintain that Liferay, even with it's issues, makes my life easier and allows me to offer my clients more value for their money by allowing me to use the tools it provides to solve their business problems rather than spend countless hours building boilerplate code.

So, you have done an execllent job of getting up onto your soap boxes and complaining about the things they don't like. Can you tell me some ot the things that you DO like about Liferay?
thumbnail
Eric COQUELIN,修改在7 年前。

RE: Is Liferay 7 GA3 ready ?

Expert 帖子: 254 加入日期: 13-11-3 最近的帖子
Andrew,

If we were unhappy with Liferay, we wouldn't have raised anything. We would have just moved to another product and that's all. But that's not the case, we still believe it is a good one (perhaps the best so far) but Liferay 7 (CE in my case) have brought a lot of frustrations. And yes, I got use to a high standard quality and I wish Liferay 7 GA3 (CE again) was as good as Liferay 6.2 GA6 (CE again) in terms of compliance with documentation, performance and bugs. Indeed, I do not pay and neither most of my clients thus I should not complain. But, we need to start somewhere like many many many companies which may subscribe later ;)

Still looking forward GA4 !

Regards,
thumbnail
David H Nebinger,修改在7 年前。

RE: Is Liferay 7 GA3 ready ?

Liferay Legend 帖子: 14914 加入日期: 06-9-2 最近的帖子
Eric COQUELIN:
And yes, I got use to a high standard quality and I wish Liferay 7 GA3 (CE again) was as good as Liferay 6.2 GA6 (CE again),



Ah, and there's the rub. 6.2 GA0 through GA5 all had a similar set of bugs. The Liferay 7 GAs will also, but that final solid GA takes awhile to get to. Liferay 7 will get to a good GA, but that only happens when folks report the bugs they find. If you're sitting on 4 pages of bugs and they're not reported, there's no guarantee they will ever be resolved.

There are some folks at Liferay who monitor the forums, but certainly not for bug reports. They have to get opened as tickets on issues.liferay.com if you have any chance of seeing a resolution.

And if the tickets are there, be sure to vote for them as that raises their visibility within Liferay.
thumbnail
Andrew Jardine,修改在7 年前。

RE: Is Liferay 7 GA3 ready ?

Liferay Legend 帖子: 2416 加入日期: 10-12-22 最近的帖子
Hey Eric,

I had no issue with your post. The main point of these forums is to solicti help which is exactly what you did. You pointed to concrete issues that raised concerns with you and asked for possible solutions. While not ALL your issues were resolved I think we were able to get a couple of them off the list -- and in some cases they had nothing to do with LR at all right? What annoys me is the "rant without reason or explanation". I'm happy to try to help solve problems --- even those that I am not familiar with because two heads are always better than one. But I am not paid to be on here, I volunteer my time as a way to try to give back to the product -- and if someone is having a shitty day (and this is not a criticism o your post) and needs an outlet to vent, this is not the place. Go get a therapist who also doesn't give a shit about what you are complaining about, but is at least paid to listen. Posts like

mmmmmm.. sounds shitty but true


don't really contribute much to the conversation, so when I see criticsm without self reflection, that really pisses me off. Like you I am sure, I have worked with countless arrogant developers who think their work is a gift to the rest of us. I'm an old fart now and have no time or patience for that anymore ;)

I too am anxiously awaiting another GA -- but in the mean time I have taken to doing nightly builds with the latest fixes -- producing my own intermediary GA builds. Is there an option for you to do that same? It might help you to push past some of your issues. If it's an option, here is a blog post to get you on the track -- https://web.liferay.com/web/kyle.stiemann/blog/-/blogs/how-can-i-build-liferay-7-0-from-source-

I am using the Tomcat bundle so it basically boils down to a git clone command and then an ant clean all.
thumbnail
David H Nebinger,修改在7 年前。

RE: Is Liferay 7 GA3 ready ?

Liferay Legend 帖子: 14914 加入日期: 06-9-2 最近的帖子
Andrew Jardine:
I too am anxiously awaiting another GA -- but in the mean time I have taken to doing nightly builds with the latest fixes -- producing my own intermediary GA builds.


Note the important thing to keep in mind w/ this (and I'm sure Andrew knows this, just pointing out for others), there is no guarantee that builds from master are going to work. There is normally a high probability at this point, but unlike the GAs nightly builds have not completed testing, only code reviews.