Github links broken

[joeygold]

https://github.com/Microchip-MPLAB-Harmony/gfx/tree/v3.8.0/apps

[MHGC]

The application repositories are still in release testing and will be published shortly.

The same is true for the Aria Archive repository.
https://github.com/Microchip-MPLAB-Harmony/gfx_aria

Thank you for your patience.

[LuisCRSousa]

The application repositories are still in release testing and will be published shortly.

The same is true for the Aria Archive repository.
https://github.com/Microchip-MPLAB-Harmony/gfx_aria

Thank you for your patience.

I don't understand how is possible to:

• Deprecate aria 1 year later when it became ready for production
• Remove aria from the main repository and the backup links don't work 1 month after the fact
• Introduce a new gfx library still in beta as the oficial library, that have a lot of problems including not working in MAC

How is this possible to happen at Microchip?

[MHGC]

Hi both Aria and Legato, as well as all the tools involved, in the Harmony Graphics Suite is supported by a small team of three engineers.

Aria has its limitations due to various architectural reasons. It is imperative to focus on Legato's development to get it to match Aria's level of support and documentation.

Is there a particular MCU32 product your are developing for?

[LuisCRSousa]

I've developed an extremely complex UI app using the PIC32MZDA and harmony gfx aria library underneath.
This app have more than 100 different types of screens, uses extensively custom widgets, multiple fonts, internacionalization, images and all aria API underneath.
It took more than 1 year to be developed (started with the release of harmony 3), is currently being certified as part of the hardware that costed out multiple times the value of my house and to be released in multiple countries with many years of support.

I'm sorry.. but I'm a bit frustated that Microchip have a great long livespan for MCUs and the tottaly opposite for it's software libraries. My product will ship with a deprecated library..

I undestand that the best bet for Microchip may be supporting Legato, but it shouldn't tottaly deprecate it's old
library so rapidally and when Legato are still not production ready.
These type of libraries should have a LTS version to keep up with the hardware at some level.

Don't get me wrong, I love microchip, we are using more than 5 types od MCUs for more than 10 products, we are very happy with them and we will continue to use them. I'm just frustated by the hamony part of things.

Maybe show my dissatisfaction as a way to get Microchip to support Harmony team.

[MHGC]

We hear and appreciate your frustration and dissatisfaction.

We were keenly aware of the impact and rippling affect when we made the difficult decision to start Legato and to start moving away from Aria, almost two years ago. We had spent over a man-decade of effort developing Aria and it was not a choice we made lightly.

Aria's architecture targets MIPS core devices such as PIC32MZ EF and PIC32MZ DA. It was not designed for the M4-core devices brought on by the Atmel merger that became a large part of the Harmony support portfolio.

Another factor that pushed the decision towards a new graphics library was that Aria's font rendering pipeline is not able to support kerning data built into the font. This has caused text/label spacing difficulties for a size-able customer base.

Another customer feedback was that Aria was over-engineered, with versatility prioritized over draw performance.

We set out to create Legato with all of the above in-mind, take as much of the good parts of Aria and rebuild the rest. Since we had limited engineering bandwidth, our idea was to co-support Aria, as Legato is being build up. We started from M4 and M0 devices and moved up in performance to LCC-devices such as MZ EF and SAM E70. We even rewrote the Legato Composer from Java to C++ to improve user experience. We are finally working on higher performance devices as of v3.7.0. We added multi-layer, double-buffer, GPU support for PIC32MZ DA in v3.8.x and are now adding that back for the MPU's SAM A5D2/SAM 9x60.

We are seeing upwards of 200% improvement in draw performance for M4/M7/MZ EF class devices. For PIC32MZ DA, we see similar improvements in font rendering, but image rendering and fills have taken a hit. We are continuing to work on that.

Ultimately, there is a sizable customer base with Aria-based products that we encourage to migrate to Legato at their convenience. I readily admit our execution has not been good. We are short in a lot of areas such as detailed documentation and migration guides.

Given you have custom widgets and well-developed application API calls, the transition is not an insignificant endeavor. But there are encouraging signs, several PIC32MZ DA customers have made the transition from Aria to Legato and reporting good results.

Once again I apologize for the inconvenience, the gfx_aria archive repo will be published this week.

No more rewrites. Legato will be the graphics library to stay for for foreseeable future.

[LuisCRSousa]

Thank you for your response.

Me and my team were gladly supporting Aria by reporting issues and sugesting features since mid last year. We knew since the begining of our project that Aria was not perfect and we even discussed to use other MCU or just other gfx library.
We decided to give it a go with Aria because our team was already developing with Microchip for a lot of years.
We even went last here to Masters at Phoenix and personally discussed some harmony gfx details with Jesse Hunter (regional marketing manager).
It was in the Masters that we discovered that Legato was being developed but we couldn't wait for it as at the time we had already started with Aria and we couldn't know when legato was being released and, most important, production ready.

We are disappointed that Aria is deprecated when we are releasing our product.
We also undestand Aria limitations and the need for a new, performance first gfx library with ARM support. I hope that the production ready version will be release as soon as possible and that Microchip supports more the GFX team, because 3 engineers is too little for the work that needs to be done.

For now we don't know if we'll upgrade our app to Legato. It's just too much work. But when legato is released for production ready we'll discuss it again internally.

Hope everything the best for Harmony team!